[clang] [RISCV] Support Xsfvqmaccdod and Xsfvqmaccqoq extensions (PR #68295)
Brandon Wu via cfe-commits
cfe-commits at lists.llvm.org
Tue Oct 10 20:03:39 PDT 2023
https://github.com/4vtomat updated https://github.com/llvm/llvm-project/pull/68295
>From 58fd3ab3fb839095be7b6e7119859cc5b18938c3 Mon Sep 17 00:00:00 2001
From: Brandon Wu <brandon.wu at sifive.com>
Date: Wed, 27 Sep 2023 08:30:19 -0700
Subject: [PATCH 1/4] [RISCV] Support Xsfvqmaccdod and Xsfvqmaccqoq extensions
SiFive Int8 Matrix Multiplication Extensions Specification
https://sifive.cdn.prismic.io/sifive/c4f0e51d-4dd3-402a-98bc-1ffad6011259_int8-matmul-spec.pdf
---
.../clang/Basic/riscv_sifive_vector.td | 24 +++
.../clang/Basic/riscv_vector_common.td | 2 +
.../clang/Support/RISCVVIntrinsicUtils.h | 19 ++-
clang/lib/Sema/SemaRISCVVectorLookup.cpp | 2 +
clang/lib/Support/RISCVVIntrinsicUtils.cpp | 10 +-
.../non-overloaded/sf_vqmacc_2x8x2.c | 47 ++++++
.../non-overloaded/sf_vqmacc_4x8x4.c | 47 ++++++
.../non-overloaded/sf_vqmaccsu_2x8x2.c | 47 ++++++
.../non-overloaded/sf_vqmaccsu_4x8x4.c | 47 ++++++
.../non-overloaded/sf_vqmaccu_2x8x2.c | 47 ++++++
.../non-overloaded/sf_vqmaccu_4x8x4.c | 47 ++++++
.../non-overloaded/sf_vqmaccus_2x8x2.c | 47 ++++++
.../non-overloaded/sf_vqmaccus_4x8x4.c | 47 ++++++
.../non-policy/overloaded/sf_vqmacc_2x8x2.c | 47 ++++++
.../non-policy/overloaded/sf_vqmacc_4x8x4.c | 47 ++++++
.../non-policy/overloaded/sf_vqmaccsu_2x8x2.c | 47 ++++++
.../non-policy/overloaded/sf_vqmaccsu_4x8x4.c | 47 ++++++
.../non-policy/overloaded/sf_vqmaccu_2x8x2.c | 47 ++++++
.../non-policy/overloaded/sf_vqmaccu_4x8x4.c | 47 ++++++
.../non-policy/overloaded/sf_vqmaccus_2x8x2.c | 47 ++++++
.../non-policy/overloaded/sf_vqmaccus_4x8x4.c | 47 ++++++
.../policy/non-overloaded/sf_vqmacc_2x8x2.c | 47 ++++++
.../policy/non-overloaded/sf_vqmacc_4x8x4.c | 47 ++++++
.../policy/non-overloaded/sf_vqmaccsu_2x8x2.c | 47 ++++++
.../policy/non-overloaded/sf_vqmaccsu_4x8x4.c | 47 ++++++
.../policy/non-overloaded/sf_vqmaccu_2x8x2.c | 47 ++++++
.../policy/non-overloaded/sf_vqmaccu_4x8x4.c | 47 ++++++
.../policy/non-overloaded/sf_vqmaccus_2x8x2.c | 47 ++++++
.../policy/non-overloaded/sf_vqmaccus_4x8x4.c | 47 ++++++
.../policy/overloaded/sf_vqmacc_2x8x2.c | 47 ++++++
.../policy/overloaded/sf_vqmacc_4x8x4.c | 47 ++++++
.../policy/overloaded/sf_vqmaccsu_2x8x2.c | 47 ++++++
.../policy/overloaded/sf_vqmaccsu_4x8x4.c | 47 ++++++
.../policy/overloaded/sf_vqmaccu_2x8x2.c | 47 ++++++
.../policy/overloaded/sf_vqmaccu_4x8x4.c | 47 ++++++
.../policy/overloaded/sf_vqmaccus_2x8x2.c | 47 ++++++
.../policy/overloaded/sf_vqmaccus_4x8x4.c | 47 ++++++
.../test/Sema/rvv-required-features-invalid.c | 8 +
clang/test/Sema/rvv-required-features.c | 11 +-
clang/utils/TableGen/RISCVVEmitter.cpp | 2 +
llvm/include/llvm/IR/IntrinsicsRISCVXsf.td | 20 +++
llvm/lib/Support/RISCVISAInfo.cpp | 6 +
.../RISCV/Disassembler/RISCVDisassembler.cpp | 4 +
llvm/lib/Target/RISCV/RISCVFeatures.td | 16 ++
llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td | 90 ++++++++++
.../test/CodeGen/RISCV/rvv/sf_vqmacc_2x8x2.ll | 159 ++++++++++++++++++
.../test/CodeGen/RISCV/rvv/sf_vqmacc_4x8x4.ll | 159 ++++++++++++++++++
.../CodeGen/RISCV/rvv/sf_vqmaccsu_2x8x2.ll | 159 ++++++++++++++++++
.../CodeGen/RISCV/rvv/sf_vqmaccsu_4x8x4.ll | 159 ++++++++++++++++++
.../CodeGen/RISCV/rvv/sf_vqmaccu_2x8x2.ll | 159 ++++++++++++++++++
.../CodeGen/RISCV/rvv/sf_vqmaccu_4x8x4.ll | 159 ++++++++++++++++++
.../CodeGen/RISCV/rvv/sf_vqmaccus_2x8x2.ll | 159 ++++++++++++++++++
.../CodeGen/RISCV/rvv/sf_vqmaccus_4x8x4.ll | 159 ++++++++++++++++++
llvm/test/MC/RISCV/rvv/xsfvqmacc.s | 57 +++++++
llvm/unittests/Support/RISCVISAInfoTest.cpp | 2 +
55 files changed, 3039 insertions(+), 10 deletions(-)
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_4x8x4.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_2x8x2.c
create mode 100644 clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_4x8x4.c
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_2x8x2.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_4x8x4.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_2x8x2.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_4x8x4.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_2x8x2.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_4x8x4.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_2x8x2.ll
create mode 100644 llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_4x8x4.ll
create mode 100644 llvm/test/MC/RISCV/rvv/xsfvqmacc.s
diff --git a/clang/include/clang/Basic/riscv_sifive_vector.td b/clang/include/clang/Basic/riscv_sifive_vector.td
index 6583a7eb7b2e59b..079dc7768fc0580 100644
--- a/clang/include/clang/Basic/riscv_sifive_vector.td
+++ b/clang/include/clang/Basic/riscv_sifive_vector.td
@@ -103,3 +103,27 @@ let SupportOverloading = false in {
defm sf_vc_v_fvw : RVVVCIXBuiltinSet<["si"], "UwKzUwUvFe", [-1, 0, 2, 3], UseGPR=0>;
}
}
+
+multiclass RVVVQMACCBuiltinSet<list<list<string>> suffixes_prototypes> {
+ let OverloadedName = NAME,
+ Name = NAME,
+ HasMasked = false,
+ Log2LMUL = [0, 1, 2, 3] in
+ defm NAME : RVVOutOp1Op2BuiltinSet<NAME, "c", suffixes_prototypes>;
+}
+
+let UnMaskedPolicyScheme = HasPolicyOperand in
+ let RequiredFeatures = ["Xsfvqmaccdod"] in {
+ defm sf_vqmaccu_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44SUvUv"]]>;
+ defm sf_vqmacc_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44Svv"]]>;
+ defm sf_vqmaccus_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44SUvv"]]>;
+ defm sf_vqmaccsu_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44SvUv"]]>;
+ }
+
+let UnMaskedPolicyScheme = HasPolicyOperand in
+ let RequiredFeatures = ["Xsfvqmaccqoq"] in {
+ defm sf_vqmaccu_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44SUvUv"]]>;
+ defm sf_vqmacc_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44Svv"]]>;
+ defm sf_vqmaccus_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44SUvv"]]>;
+ defm sf_vqmaccsu_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44SvUv"]]>;
+ }
diff --git a/clang/include/clang/Basic/riscv_vector_common.td b/clang/include/clang/Basic/riscv_vector_common.td
index 141fac9d68e6d54..b758b7c13b37ec1 100644
--- a/clang/include/clang/Basic/riscv_vector_common.td
+++ b/clang/include/clang/Basic/riscv_vector_common.td
@@ -59,6 +59,8 @@
// element type which is four times as wide as the element type of 'v'
// o: computes a vector type identical to what 'v' computes except for the
// element type which is eight times as wide as the element type of 'v'
+// 4: computes a vector type with LMUL identical to what 'v' computes except
+// for the SEW which is 4 times as wide as the SEW of 'v'
// m: computes a vector type identical to what 'v' computes except for the
// element type which is bool
// 0: void type, ignores "t"
diff --git a/clang/include/clang/Support/RISCVVIntrinsicUtils.h b/clang/include/clang/Support/RISCVVIntrinsicUtils.h
index 8ba57d77221dc52..cb76ef598c8ad0c 100644
--- a/clang/include/clang/Support/RISCVVIntrinsicUtils.h
+++ b/clang/include/clang/Support/RISCVVIntrinsicUtils.h
@@ -35,6 +35,7 @@ enum class VectorTypeModifier : uint8_t {
Widening2XVector,
Widening4XVector,
Widening8XVector,
+ SEW4XVector,
MaskVector,
Log2EEW3,
Log2EEW4,
@@ -485,14 +486,16 @@ enum RVVRequire : uint16_t {
RVV_REQ_RV64 = 1 << 0,
RVV_REQ_ZvfhminOrZvfh = 1 << 1,
RVV_REQ_Xsfvcp = 1 << 2,
- RVV_REQ_Zvbb = 1 << 3,
- RVV_REQ_Zvbc = 1 << 4,
- RVV_REQ_Zvkb = 1 << 5,
- RVV_REQ_Zvkg = 1 << 6,
- RVV_REQ_Zvkned = 1 << 7,
- RVV_REQ_Zvknha = 1 << 8,
- RVV_REQ_Zvksed = 1 << 9,
- RVV_REQ_Zvksh = 1 << 10,
+ RVV_REQ_Xsfvqmaccdod = 1 << 3,
+ RVV_REQ_Xsfvqmaccqoq = 1 << 4,
+ RVV_REQ_Zvbb = 1 << 5,
+ RVV_REQ_Zvbc = 1 << 6,
+ RVV_REQ_Zvkb = 1 << 7,
+ RVV_REQ_Zvkg = 1 << 8,
+ RVV_REQ_Zvkned = 1 << 9,
+ RVV_REQ_Zvknha = 1 << 10,
+ RVV_REQ_Zvksed = 1 << 11,
+ RVV_REQ_Zvksh = 1 << 12,
LLVM_MARK_AS_BITMASK_ENUM(RVV_REQ_Zvksh)
};
diff --git a/clang/lib/Sema/SemaRISCVVectorLookup.cpp b/clang/lib/Sema/SemaRISCVVectorLookup.cpp
index ae584dc68719901..acdf3260007bc3a 100644
--- a/clang/lib/Sema/SemaRISCVVectorLookup.cpp
+++ b/clang/lib/Sema/SemaRISCVVectorLookup.cpp
@@ -205,6 +205,8 @@ void RISCVIntrinsicManagerImpl::ConstructRVVIntrinsics(
static const std::pair<const char *, RVVRequire> FeatureCheckList[] = {
{"64bit", RVV_REQ_RV64},
{"xsfvcp", RVV_REQ_Xsfvcp},
+ {"xsfvqmaccdod", RVV_REQ_Xsfvqmaccdod},
+ {"xsfvqmaccqoq", RVV_REQ_Xsfvqmaccqoq},
{"experimental-zvbb", RVV_REQ_Zvbb},
{"experimental-zvbc", RVV_REQ_Zvbc},
{"experimental-zvkb", RVV_REQ_Zvkb},
diff --git a/clang/lib/Support/RISCVVIntrinsicUtils.cpp b/clang/lib/Support/RISCVVIntrinsicUtils.cpp
index c105db434dc43c9..ce46a3f5c662e87 100644
--- a/clang/lib/Support/RISCVVIntrinsicUtils.cpp
+++ b/clang/lib/Support/RISCVVIntrinsicUtils.cpp
@@ -411,6 +411,10 @@ PrototypeDescriptor::parsePrototypeDescriptor(
PT = BaseTypeModifier::Vector;
VTM = VectorTypeModifier::Widening8XVector;
break;
+ case '4':
+ PT = BaseTypeModifier::Vector;
+ VTM = VectorTypeModifier::SEW4XVector;
+ break;
case 'm':
PT = BaseTypeModifier::Vector;
VTM = VectorTypeModifier::MaskVector;
@@ -687,6 +691,10 @@ void RVVType::applyModifier(const PrototypeDescriptor &Transformer) {
LMUL.MulLog2LMUL(3);
Scale = LMUL.getScale(ElementBitwidth);
break;
+ case VectorTypeModifier::SEW4XVector:
+ ElementBitwidth *= 4;
+ Scale = LMUL.getScale(ElementBitwidth);
+ break;
case VectorTypeModifier::MaskVector:
ScalarType = ScalarTypeKind::Boolean;
Scale = LMUL.getScale(ElementBitwidth);
@@ -1150,7 +1158,7 @@ void RVVIntrinsic::updateNamesAndPolicy(
SmallVector<PrototypeDescriptor> parsePrototypes(StringRef Prototypes) {
SmallVector<PrototypeDescriptor> PrototypeDescriptors;
- const StringRef Primaries("evwqom0ztul");
+ const StringRef Primaries("evwqo4m0ztul");
while (!Prototypes.empty()) {
size_t Idx = 0;
// Skip over complex prototype because it could contain primitive type
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_2x8x2.c
new file mode 100644
index 000000000000000..29ebeea7a04e089
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_2x8x2_i32m1(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_2x8x2_i32m2(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_2x8x2_i32m4(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_2x8x2_i32m8(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_4x8x4.c
new file mode 100644
index 000000000000000..935cb2e007d3564
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmacc_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_4x8x4_i32m1(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_4x8x4_i32m2(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_4x8x4_i32m4(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_4x8x4_i32m8(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_2x8x2.c
new file mode 100644
index 000000000000000..a1f95d2c150f29d
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_2x8x2_i32m1(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_2x8x2_i32m2(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_2x8x2_i32m4(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_2x8x2_i32m8(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_4x8x4.c
new file mode 100644
index 000000000000000..f34517b24bcf205
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccsu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_4x8x4_i32m1(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_4x8x4_i32m2(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_4x8x4_i32m4(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_4x8x4_i32m8(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_2x8x2.c
new file mode 100644
index 000000000000000..9f887562e12b435
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_2x8x2_i32m1(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_2x8x2_i32m2(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_2x8x2_i32m4(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_2x8x2_i32m8(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_4x8x4.c
new file mode 100644
index 000000000000000..ab7f6627ad1fb04
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_4x8x4_i32m1(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_4x8x4_i32m2(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_4x8x4_i32m4(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_4x8x4_i32m8(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_2x8x2.c
new file mode 100644
index 000000000000000..2921228700d2b85
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_2x8x2_i32m1(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_2x8x2_i32m2(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_2x8x2_i32m4(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_2x8x2_i32m8(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_4x8x4.c
new file mode 100644
index 000000000000000..d0bcdcbf40cc324
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/sf_vqmaccus_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_4x8x4_i32m1(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m1(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_4x8x4_i32m2(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_4x8x4_i32m4(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_4x8x4_i32m8(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m8(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_2x8x2.c
new file mode 100644
index 000000000000000..bef04c570710c43
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_2x8x2_i32m1(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_2x8x2_i32m2(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_2x8x2_i32m4(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_2x8x2_i32m8(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_4x8x4.c
new file mode 100644
index 000000000000000..839d09c4f9a986a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmacc_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_4x8x4_i32m1(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_4x8x4_i32m2(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_4x8x4_i32m4(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_4x8x4_i32m8(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_2x8x2.c
new file mode 100644
index 000000000000000..da0f632a3a9f5b8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_2x8x2_i32m1(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_2x8x2_i32m2(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_2x8x2_i32m4(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_2x8x2_i32m8(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_4x8x4.c
new file mode 100644
index 000000000000000..b18853043e9249a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccsu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_4x8x4_i32m1(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_4x8x4_i32m2(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_4x8x4_i32m4(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_4x8x4_i32m8(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_2x8x2.c
new file mode 100644
index 000000000000000..c58182cf3c08ef3
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_2x8x2_i32m1(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_2x8x2_i32m2(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_2x8x2_i32m4(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_2x8x2_i32m8(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_4x8x4.c
new file mode 100644
index 000000000000000..4cb966b08f23701
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_4x8x4_i32m1(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_4x8x4_i32m2(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_4x8x4_i32m4(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_4x8x4_i32m8(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_2x8x2.c
new file mode 100644
index 000000000000000..a16f576e539db33
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_2x8x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_2x8x2_i32m1(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_2x8x2_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_2x8x2_i32m2(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_2x8x2_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_2x8x2_i32m4(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_2x8x2_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_2x8x2_i32m8(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_4x8x4.c
new file mode 100644
index 000000000000000..f558151f88a3f2a
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/sf_vqmaccus_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_4x8x4_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_4x8x4_i32m1(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_4x8x4_i32m2
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_4x8x4_i32m2(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_4x8x4_i32m4
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_4x8x4_i32m4(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_4x8x4_i32m8
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_4x8x4_i32m8(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_2x8x2.c
new file mode 100644
index 000000000000000..b4f0729d2fddeb0
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_2x8x2_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_2x8x2_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_2x8x2_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_2x8x2_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_4x8x4.c
new file mode 100644
index 000000000000000..05c10840cabfaa5
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmacc_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_4x8x4_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_4x8x4_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_4x8x4_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_4x8x4_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_2x8x2.c
new file mode 100644
index 000000000000000..612f8c3cfa613ca
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_2x8x2_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_2x8x2_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_2x8x2_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_2x8x2_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_4x8x4.c
new file mode 100644
index 000000000000000..bce1a4e9443fd92
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccsu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_4x8x4_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_4x8x4_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_4x8x4_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_4x8x4_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_2x8x2.c
new file mode 100644
index 000000000000000..6ffe84df87c2cd7
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_2x8x2_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_2x8x2_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_2x8x2_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_2x8x2_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_4x8x4.c
new file mode 100644
index 000000000000000..36aaae9caebf60b
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_4x8x4_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_4x8x4_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_4x8x4_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_4x8x4_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_2x8x2.c
new file mode 100644
index 000000000000000..ebfd33f1b67a991
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_2x8x2_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_2x8x2_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_2x8x2_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_2x8x2_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_4x8x4.c
new file mode 100644
index 000000000000000..f5ac2bf0f1f3a75
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/sf_vqmaccus_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_4x8x4_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m1_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_4x8x4_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_4x8x4_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_4x8x4_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_i32m8_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_2x8x2.c
new file mode 100644
index 000000000000000..6d7ca5ab4ee79f2
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_2x8x2_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_2x8x2_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_2x8x2_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_2x8x2_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_2x8x2_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_4x8x4.c
new file mode 100644
index 000000000000000..531bc2b2b942535
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmacc_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmacc_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmacc_4x8x4_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmacc_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmacc_4x8x4_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmacc_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmacc_4x8x4_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmacc_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmacc_4x8x4_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmacc_4x8x4_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_2x8x2.c
new file mode 100644
index 000000000000000..ba6dae595bd340c
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_2x8x2_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_2x8x2_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_2x8x2_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_2x8x2_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_2x8x2_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_4x8x4.c
new file mode 100644
index 000000000000000..23bba523aaa4428
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccsu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccsu_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccsu_4x8x4_i32m1_tu(vint32m1_t vd, vint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccsu_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccsu_4x8x4_i32m2_tu(vint32m2_t vd, vint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccsu_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccsu_4x8x4_i32m4_tu(vint32m4_t vd, vint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccsu_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccsu_4x8x4_i32m8_tu(vint32m8_t vd, vint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccsu_4x8x4_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_2x8x2.c
new file mode 100644
index 000000000000000..9e15b053da244c4
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_2x8x2_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_2x8x2_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_2x8x2_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_2x8x2_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_2x8x2_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_4x8x4.c
new file mode 100644
index 000000000000000..950688c6c7851a8
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccu_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccu_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccu_4x8x4_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vuint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccu_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccu_4x8x4_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vuint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccu_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccu_4x8x4_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vuint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccu_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccu_4x8x4_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vuint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccu_4x8x4_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_2x8x2.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_2x8x2.c
new file mode 100644
index 000000000000000..3108560c5be45ea
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_2x8x2.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccdod \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_2x8x2_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_2x8x2_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_2x8x2_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_2x8x2_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_2x8x2_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_2x8x2_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_2x8x2_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_2x8x2_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_2x8x2_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_4x8x4.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_4x8x4.c
new file mode 100644
index 000000000000000..7bdce95043ee4ec
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/sf_vqmaccus_4x8x4.c
@@ -0,0 +1,47 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvqmaccqoq \
+// RUN: -disable-O0-optnone -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN: FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <sifive_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_sf_vqmaccus_4x8x4_i32m1_tu
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 8 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8.i64(<vscale x 2 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 8 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 2 x i32> [[TMP0]]
+//
+vint32m1_t test_sf_vqmaccus_4x8x4_i32m1_tu(vint32m1_t vd, vuint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x i32> @test_sf_vqmaccus_4x8x4_i32m2_tu
+// CHECK-RV64-SAME: (<vscale x 4 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 16 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8.i64(<vscale x 4 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 16 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 4 x i32> [[TMP0]]
+//
+vint32m2_t test_sf_vqmaccus_4x8x4_i32m2_tu(vint32m2_t vd, vuint8m1_t vs1, vint8m2_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x i32> @test_sf_vqmaccus_4x8x4_i32m4_tu
+// CHECK-RV64-SAME: (<vscale x 8 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 32 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8.i64(<vscale x 8 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 32 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 8 x i32> [[TMP0]]
+//
+vint32m4_t test_sf_vqmaccus_4x8x4_i32m4_tu(vint32m4_t vd, vuint8m1_t vs1, vint8m4_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_tu(vd, vs1, vs2, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x i32> @test_sf_vqmaccus_4x8x4_i32m8_tu
+// CHECK-RV64-SAME: (<vscale x 16 x i32> [[VD:%.*]], <vscale x 8 x i8> [[VS1:%.*]], <vscale x 64 x i8> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT: entry:
+// CHECK-RV64-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8.i64(<vscale x 16 x i32> [[VD]], <vscale x 8 x i8> [[VS1]], <vscale x 64 x i8> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT: ret <vscale x 16 x i32> [[TMP0]]
+//
+vint32m8_t test_sf_vqmaccus_4x8x4_i32m8_tu(vint32m8_t vd, vuint8m1_t vs1, vint8m8_t vs2, size_t vl) {
+ return __riscv_sf_vqmaccus_4x8x4_tu(vd, vs1, vs2, vl);
+}
diff --git a/clang/test/Sema/rvv-required-features-invalid.c b/clang/test/Sema/rvv-required-features-invalid.c
index 0d0d00764a31e39..56f201a507ef7c0 100644
--- a/clang/test/Sema/rvv-required-features-invalid.c
+++ b/clang/test/Sema/rvv-required-features-invalid.c
@@ -15,3 +15,11 @@ void test_vsoxei64_v_i8m1(int8_t *base, vuint64m8_t bindex, vint8m1_t value, siz
void test_xsfvcp_sf_vc_x_se_u64m1(uint64_t rs1, size_t vl) {
__riscv_sf_vc_x_se_u64m1(1, 1, 1, rs1, vl); // expected-error {{call to undeclared function '__riscv_sf_vc_x_se_u64m1'}}
}
+
+void test_xsfvqmaccdod() {
+ __riscv_sf_vqmacc_2x8x2(); // expected-error {{call to undeclared function '__riscv_sf_vqmacc_2x8x2'}}
+}
+
+void test_xsfvqmaccqoq() {
+ __riscv_sf_vqmacc_4x8x4(); // expected-error {{call to undeclared function '__riscv_sf_vqmacc_4x8x4'}}
+}
diff --git a/clang/test/Sema/rvv-required-features.c b/clang/test/Sema/rvv-required-features.c
index c3b7965599e68fb..ed665c8ecbd4927 100644
--- a/clang/test/Sema/rvv-required-features.c
+++ b/clang/test/Sema/rvv-required-features.c
@@ -1,5 +1,6 @@
// REQUIRES: riscv-registered-target
-// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvcp %s -fsyntax-only -verify
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +xsfvcp \
+// RUN: -target-feature +xsfvqmaccdod -target-feature +xsfvqmaccqoq %s -fsyntax-only -verify
// expected-no-diagnostics
@@ -17,3 +18,11 @@ void test_vsoxei64_v_i8m1(int8_t *base, vuint64m8_t bindex, vint8m1_t value, siz
void test_sf_vc_x_se_u64m1(uint64_t rs1, size_t vl) {
__riscv_sf_vc_x_se_u64m1(1, 1, 1, rs1, vl);
}
+
+void test_xsfvqmaccdod(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ __riscv_sf_vqmacc_2x8x2(vd, vs1, vs2, vl);
+}
+
+void test_xsfvqmaccqoq(vint32m1_t vd, vint8m1_t vs1, vint8m1_t vs2, size_t vl) {
+ __riscv_sf_vqmacc_4x8x4(vd, vs1, vs2, vl);
+}
diff --git a/clang/utils/TableGen/RISCVVEmitter.cpp b/clang/utils/TableGen/RISCVVEmitter.cpp
index 41025926058ed07..06abe4ce2e3941d 100644
--- a/clang/utils/TableGen/RISCVVEmitter.cpp
+++ b/clang/utils/TableGen/RISCVVEmitter.cpp
@@ -656,6 +656,8 @@ void RVVEmitter::createRVVIntrinsics(
.Case("RV64", RVV_REQ_RV64)
.Case("ZvfhminOrZvfh", RVV_REQ_ZvfhminOrZvfh)
.Case("Xsfvcp", RVV_REQ_Xsfvcp)
+ .Case("Xsfvqmaccdod", RVV_REQ_Xsfvqmaccdod)
+ .Case("Xsfvqmaccqoq", RVV_REQ_Xsfvqmaccqoq)
.Case("Zvbb", RVV_REQ_Zvbb)
.Case("Zvbc", RVV_REQ_Zvbc)
.Case("Zvkb", RVV_REQ_Zvkb)
diff --git a/llvm/include/llvm/IR/IntrinsicsRISCVXsf.td b/llvm/include/llvm/IR/IntrinsicsRISCVXsf.td
index c8d24ec7d83addf..4381c55a3f91dc8 100644
--- a/llvm/include/llvm/IR/IntrinsicsRISCVXsf.td
+++ b/llvm/include/llvm/IR/IntrinsicsRISCVXsf.td
@@ -128,8 +128,28 @@ let TargetPrefix = "riscv" in {
}
}
+ class RISCVSFCustomVMACC
+ : DefaultAttrsIntrinsic< [llvm_anyvector_ty],
+ [LLVMMatchType<0>, llvm_anyvector_ty, llvm_anyvector_ty,
+ llvm_anyint_ty, LLVMMatchType<3>],
+ [ImmArg<ArgIndex<4>>, IntrNoMem] >, RISCVVIntrinsic {
+ let VLOperand = 3;
+ }
+
defm "" : RISCVSFCustomVC_X<["x", "i"]>;
defm "" : RISCVSFCustomVC_XV<["x", "i", "v", "f"]>;
defm "" : RISCVSFCustomVC_XVV<["x", "i", "v", "f"]>;
defm "" : RISCVSFCustomVC_XVW<["x", "i", "v", "f"]>;
+
+ // XSfvqmaccdod
+ def int_riscv_sf_vqmaccu_2x8x2 : RISCVSFCustomVMACC;
+ def int_riscv_sf_vqmacc_2x8x2 : RISCVSFCustomVMACC;
+ def int_riscv_sf_vqmaccus_2x8x2 : RISCVSFCustomVMACC;
+ def int_riscv_sf_vqmaccsu_2x8x2 : RISCVSFCustomVMACC;
+
+ // XSfvqmaccqoq
+ def int_riscv_sf_vqmaccu_4x8x4 : RISCVSFCustomVMACC;
+ def int_riscv_sf_vqmacc_4x8x4 : RISCVSFCustomVMACC;
+ def int_riscv_sf_vqmaccus_4x8x4 : RISCVSFCustomVMACC;
+ def int_riscv_sf_vqmaccsu_4x8x4 : RISCVSFCustomVMACC;
} // TargetPrefix = "riscv"
diff --git a/llvm/lib/Support/RISCVISAInfo.cpp b/llvm/lib/Support/RISCVISAInfo.cpp
index 72d33e1e65c8f58..22208e2e0c2950b 100644
--- a/llvm/lib/Support/RISCVISAInfo.cpp
+++ b/llvm/lib/Support/RISCVISAInfo.cpp
@@ -73,6 +73,8 @@ static const RISCVSupportedExtension SupportedExtensions[] = {
{"xcvsimd", RISCVExtensionVersion{1, 0}},
{"xsfcie", RISCVExtensionVersion{1, 0}},
{"xsfvcp", RISCVExtensionVersion{1, 0}},
+ {"xsfvqmaccdod", RISCVExtensionVersion{1, 0}},
+ {"xsfvqmaccqoq", RISCVExtensionVersion{1, 0}},
{"xtheadba", RISCVExtensionVersion{1, 0}},
{"xtheadbb", RISCVExtensionVersion{1, 0}},
{"xtheadbs", RISCVExtensionVersion{1, 0}},
@@ -991,6 +993,8 @@ static const char *ImpliedExtsF[] = {"zicsr"};
static const char *ImpliedExtsV[] = {"zvl128b", "zve64d"};
static const char *ImpliedExtsXTHeadVdot[] = {"v"};
static const char *ImpliedExtsXsfvcp[] = {"zve32x"};
+static const char *ImpliedExtsXsfvqmaccdod[] = {"zve32x"};
+static const char *ImpliedExtsXsfvqmaccqoq[] = {"zve32x"};
static const char *ImpliedExtsZacas[] = {"a"};
static const char *ImpliedExtsZcb[] = {"zca"};
static const char *ImpliedExtsZcd[] = {"d", "zca"};
@@ -1058,6 +1062,8 @@ static constexpr ImpliedExtsEntry ImpliedExts[] = {
{{"f"}, {ImpliedExtsF}},
{{"v"}, {ImpliedExtsV}},
{{"xsfvcp"}, {ImpliedExtsXsfvcp}},
+ {{"xsfvqmaccdod"}, {ImpliedExtsXsfvqmaccdod}},
+ {{"xsfvqmaccqoq"}, {ImpliedExtsXsfvqmaccqoq}},
{{"xtheadvdot"}, {ImpliedExtsXTHeadVdot}},
{{"zacas"}, {ImpliedExtsZacas}},
{{"zcb"}, {ImpliedExtsZcb}},
diff --git a/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp b/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp
index d561d90d3088c1a..845d23945b1d4df 100644
--- a/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp
+++ b/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp
@@ -558,6 +558,10 @@ DecodeStatus RISCVDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
"XTHeadVdot custom opcode table");
TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfvcp, DecoderTableXSfvcp32,
"SiFive VCIX custom opcode table");
+ TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfvqmaccdod, DecoderTableXSfvqmaccdod32,
+ "SiFive Matrix Multiplication Instruction opcode table");
+ TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfvqmaccqoq, DecoderTableXSfvqmaccqoq32,
+ "SiFive Matrix Multiplication Instruction opcode table");
TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfcie, DecoderTableXSfcie32,
"Sifive CIE custom opcode table");
TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXCVbitmanip,
diff --git a/llvm/lib/Target/RISCV/RISCVFeatures.td b/llvm/lib/Target/RISCV/RISCVFeatures.td
index 3d3486b7fa89563..cdee139c587444f 100644
--- a/llvm/lib/Target/RISCV/RISCVFeatures.td
+++ b/llvm/lib/Target/RISCV/RISCVFeatures.td
@@ -813,6 +813,22 @@ def HasVendorXSfcie : Predicate<"Subtarget->hasVendorXSfcie()">,
AssemblerPredicate<(all_of FeatureVendorXSfcie),
"'XSfcie' (SiFive Custom Instruction Extension SCIE.)">;
+def FeatureVendorXSfvqmaccdod
+ : SubtargetFeature<"xsfvqmaccdod", "HasVendorXSfvqmaccdod", "true",
+ "'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2))",
+ [FeatureStdExtZve32x]>;
+def HasVendorXSfvqmaccdod : Predicate<"Subtarget->hasVendorXSfvqmaccdod()">,
+ AssemblerPredicate<(all_of FeatureVendorXSfvqmaccdod),
+ "'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2))">;
+
+def FeatureVendorXSfvqmaccqoq
+ : SubtargetFeature<"xsfvqmaccqoq", "HasVendorXSfvqmaccqoq", "true",
+ "'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))",
+ [FeatureStdExtZve32x]>;
+def HasVendorXSfvqmaccqoq : Predicate<"Subtarget->hasVendorXSfvqmaccqoq()">,
+ AssemblerPredicate<(all_of FeatureVendorXSfvqmaccqoq),
+ "'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))">;
+
def FeatureVendorXCVbitmanip
: SubtargetFeature<"xcvbitmanip", "HasVendorXCVbitmanip", "true",
"'XCVbitmanip' (CORE-V Bit Manipulation)">;
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td b/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
index 3975b8426256ac7..111358cb0fe080b 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
@@ -178,6 +178,19 @@ multiclass CustomSiFiveVCIX<string suffix, VCIXType type,
InTyRs1, 1>;
}
+let hasSideEffects = 0, mayLoad = 0, mayStore = 0 in {
+class CustomSiFiveVMACC<bits<6> funct6, RISCVVFormat opv, string opcodestr>
+ : RVInstVCCustom2<funct6{5-2}, opv.Value, (outs VR:$rd), (ins VR:$rs1, VR:$rs2),
+ opcodestr, "$rd, $rs1, $rs2"> {
+ let vm = 1;
+ let funct6_lo2 = funct6{1-0};
+}
+}
+
+multiclass CustomSiFiveVMACC<bits<6> funct6, RISCVVFormat opv, string opcodestr> {
+ def _VV : CustomSiFiveVMACC<funct6, opv, opcodestr>;
+}
+
let Predicates = [HasVendorXSfvcp], mayLoad = 0, mayStore = 0,
hasSideEffects = 1, hasNoSchedulingInfo = 1, DecoderNamespace = "XSfvcp" in {
defm X : CustomSiFiveVCIX<"x", VCIX_X, uimm5, uimm5, GPR>, Sched<[]>;
@@ -196,6 +209,20 @@ let Predicates = [HasVendorXSfvcp], mayLoad = 0, mayStore = 0,
defm FVW : CustomSiFiveVCIX<"fvw", VCIX_XVW, VR, VR, FPR32>, Sched<[]>;
}
+let Predicates = [HasVendorXSfvqmaccdod], DecoderNamespace = "XSfvqmaccdod" in {
+ defm VQMACCU_2x8x2 : CustomSiFiveVMACC<0b101100, OPMVV, "sf.vqmaccu.2x8x2">;
+ defm VQMACC_2x8x2 : CustomSiFiveVMACC<0b101101, OPMVV, "sf.vqmacc.2x8x2">;
+ defm VQMACCUS_2x8x2 : CustomSiFiveVMACC<0b101110, OPMVV, "sf.vqmaccus.2x8x2">;
+ defm VQMACCSU_2x8x2 : CustomSiFiveVMACC<0b101111, OPMVV, "sf.vqmaccsu.2x8x2">;
+}
+
+let Predicates = [HasVendorXSfvqmaccqoq], DecoderNamespace = "XSfvqmaccqoq" in {
+ defm VQMACCU_4x8x4 : CustomSiFiveVMACC<0b111100, OPMVV, "sf.vqmaccu.4x8x4">;
+ defm VQMACC_4x8x4 : CustomSiFiveVMACC<0b111101, OPMVV, "sf.vqmacc.4x8x4">;
+ defm VQMACCUS_4x8x4 : CustomSiFiveVMACC<0b111110, OPMVV, "sf.vqmaccus.4x8x4">;
+ defm VQMACCSU_4x8x4 : CustomSiFiveVMACC<0b111111, OPMVV, "sf.vqmaccsu.4x8x4">;
+}
+
class VPseudoVC_X<Operand OpClass, DAGOperand RS1Class,
bit HasSideEffect = 1> :
Pseudo<(outs),
@@ -318,6 +345,16 @@ multiclass VPseudoVC_XVW<LMULInfo m, DAGOperand RS1Class,
}
}
+multiclass VPseudoSiFiveVMACC<string mx, VReg vd_type, VReg vs2_type,
+ string Constraint = ""> {
+ def "Pseudo" # NAME # "_VV_" # mx : VPseudoTernaryNoMaskWithPolicy<vd_type, V_M1.vrclass, vs2_type, Constraint>;
+}
+
+multiclass VPseudoSiFiveVQMACC<string Constraint = ""> {
+ foreach m = MxListVF8 in
+ defm NAME : VPseudoSiFiveVMACC<m.MX, m.vrclass, m.vrclass, Constraint>;
+}
+
let Predicates = [HasVendorXSfvcp] in {
foreach m = MxList in {
defm X : VPseudoVC_X<m, GPR>;
@@ -346,6 +383,20 @@ let Predicates = [HasVendorXSfvcp] in {
}
}
+let Predicates = [HasVendorXSfvqmaccdod], DecoderNamespace = "XSfvqmaccdod" in {
+ defm VQMACCU_2x8x2 : VPseudoSiFiveVQMACC;
+ defm VQMACC_2x8x2 : VPseudoSiFiveVQMACC;
+ defm VQMACCUS_2x8x2 : VPseudoSiFiveVQMACC;
+ defm VQMACCSU_2x8x2 : VPseudoSiFiveVQMACC;
+}
+
+let Predicates = [HasVendorXSfvqmaccqoq], DecoderNamespace = "XSfvqmaccqoq" in {
+ defm VQMACCU_4x8x4 : VPseudoSiFiveVQMACC;
+ defm VQMACC_4x8x4 : VPseudoSiFiveVQMACC;
+ defm VQMACCUS_4x8x4 : VPseudoSiFiveVQMACC;
+ defm VQMACCSU_4x8x4 : VPseudoSiFiveVQMACC;
+}
+
class VPatVC_OP4<string intrinsic_name,
string inst,
ValueType op2_type,
@@ -476,6 +527,31 @@ class GetFTypeInfo<int Sew> {
!eq(Scalar, f64) : "FPR64");
}
+multiclass VPatVMACC<string intrinsic, string instruction,
+ list<VTypeInfoToWide> info_pairs, ValueType vec_m1> {
+ foreach pair = info_pairs in {
+ defvar VdInfo = pair.Wti;
+ defvar Vs2Info = pair.Vti;
+ let Predicates = [HasVInstructions] in
+ def : VPatTernaryNoMaskWithPolicy<"int_riscv_sf_" # intrinsic,
+ "Pseudo" # instruction, "VV", VdInfo.Vector,
+ vec_m1, Vs2Info.Vector,
+ Vs2Info.Log2SEW, Vs2Info.LMul,
+ VdInfo.RegClass, VR, Vs2Info.RegClass>;
+ }
+}
+
+defset list<VTypeInfoToWide> VQMACCInfoPairs = {
+ def : VTypeInfoToWide<VI8M1, VI32M1>;
+ def : VTypeInfoToWide<VI8M2, VI32M2>;
+ def : VTypeInfoToWide<VI8M4, VI32M4>;
+ def : VTypeInfoToWide<VI8M8, VI32M8>;
+}
+
+multiclass VPatVQMACC<string intrinsic, string instruction> {
+ defm : VPatVMACC<intrinsic, instruction, VQMACCInfoPairs, vint8m1_t>;
+}
+
let Predicates = [HasVendorXSfvcp] in {
foreach vti = AllVectors in {
defm : VPatVC_X<"x", "X", vti, XLenVT, GPR>;
@@ -511,6 +587,20 @@ let Predicates = [HasVendorXSfvcp] in {
}
}
+let Predicates = [HasVendorXSfvqmaccdod] in {
+ defm : VPatVQMACC<"vqmaccu_2x8x2", "VQMACCU_2x8x2">;
+ defm : VPatVQMACC<"vqmacc_2x8x2", "VQMACC_2x8x2">;
+ defm : VPatVQMACC<"vqmaccus_2x8x2", "VQMACCUS_2x8x2">;
+ defm : VPatVQMACC<"vqmaccsu_2x8x2", "VQMACCSU_2x8x2">;
+}
+
+let Predicates = [HasVendorXSfvqmaccqoq] in {
+ defm : VPatVQMACC<"vqmaccu_4x8x4", "VQMACCU_4x8x4">;
+ defm : VPatVQMACC<"vqmacc_4x8x4", "VQMACC_4x8x4">;
+ defm : VPatVQMACC<"vqmaccus_4x8x4", "VQMACCUS_4x8x4">;
+ defm : VPatVQMACC<"vqmaccsu_4x8x4", "VQMACCSU_4x8x4">;
+}
+
let Predicates = [HasVendorXSfcie] in {
let hasSideEffects = 1, mayLoad = 0, mayStore = 0, DecoderNamespace = "XSfcie" in {
def SF_CFLUSH_D_L1 : RVInstI<0b000, OPC_SYSTEM, (outs), (ins GPR:$rs1), "cflush.d.l1","$rs1">,
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_2x8x2.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_2x8x2.ll
new file mode 100644
index 000000000000000..0a3623c236486c5
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_2x8x2.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmacc_2x8x2_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmacc_2x8x2_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmacc_2x8x2_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmacc_2x8x2_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmacc_2x8x2_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmacc_2x8x2_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmacc_2x8x2_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmacc_2x8x2_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_2x8x2_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_4x8x4.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_4x8x4.ll
new file mode 100644
index 000000000000000..843e4bda4d123f7
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmacc_4x8x4.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmacc_4x8x4_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmacc_4x8x4_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmacc_4x8x4_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmacc_4x8x4_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmacc_4x8x4_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmacc_4x8x4_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmacc_4x8x4_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmacc_4x8x4_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmacc_4x8x4_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmacc.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmacc.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_2x8x2.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_2x8x2.ll
new file mode 100644
index 000000000000000..106d3183991c79e
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_2x8x2.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmaccsu_2x8x2_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmaccsu_2x8x2_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmaccsu_2x8x2_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmaccsu_2x8x2_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmaccsu_2x8x2_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmaccsu_2x8x2_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmaccsu_2x8x2_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmaccsu_2x8x2_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_2x8x2_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_4x8x4.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_4x8x4.ll
new file mode 100644
index 000000000000000..45a3b22d9618a51
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccsu_4x8x4.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmaccsu_4x8x4_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmaccsu_4x8x4_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmaccsu_4x8x4_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmaccsu_4x8x4_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmaccsu_4x8x4_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmaccsu_4x8x4_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmaccsu_4x8x4_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmaccsu_4x8x4_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccsu_4x8x4_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccsu.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccsu.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_2x8x2.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_2x8x2.ll
new file mode 100644
index 000000000000000..d9ab4559eac6904
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_2x8x2.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmaccu_2x8x2_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmaccu_2x8x2_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmaccu_2x8x2_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmaccu_2x8x2_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmaccu_2x8x2_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmaccu_2x8x2_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmaccu_2x8x2_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmaccu_2x8x2_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_2x8x2_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_4x8x4.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_4x8x4.ll
new file mode 100644
index 000000000000000..9d15ab68a091a16
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccu_4x8x4.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmaccu_4x8x4_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmaccu_4x8x4_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmaccu_4x8x4_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmaccu_4x8x4_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmaccu_4x8x4_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmaccu_4x8x4_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmaccu_4x8x4_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmaccu_4x8x4_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccu_4x8x4_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccu.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccu.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_2x8x2.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_2x8x2.ll
new file mode 100644
index 000000000000000..7ce486993c7d1c2
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_2x8x2.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccdod \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmaccus_2x8x2_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmaccus_2x8x2_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmaccus_2x8x2_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmaccus_2x8x2_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmaccus_2x8x2_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmaccus_2x8x2_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmaccus_2x8x2_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmaccus_2x8x2_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_2x8x2_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.2x8x2 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.2x8x2.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_4x8x4.ll b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_4x8x4.ll
new file mode 100644
index 000000000000000..d9b62d8a57691ac
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/sf_vqmaccus_4x8x4.ll
@@ -0,0 +1,159 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v,+xsfvqmaccqoq \
+; RUN: -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK
+
+declare <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 8 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 2 x i32> @intrinsic_vqmaccus_4x8x4_tu_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_tu_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 2 x i32> %a
+}
+
+define <vscale x 2 x i32> @intrinsic_vqmaccus_4x8x4_ta_i32m1(<vscale x 2 x i32> %0, <vscale x 8 x i8> %1, <vscale x 8 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_ta_i32m1:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v9, v10
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 2 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv2i32.nxv8i8.nxv8i8(
+ <vscale x 2 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 8 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 16 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 4 x i32> @intrinsic_vqmaccus_4x8x4_tu_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_tu_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 4 x i32> %a
+}
+
+define <vscale x 4 x i32> @intrinsic_vqmaccus_4x8x4_ta_i32m2(<vscale x 4 x i32> %0, <vscale x 8 x i8> %1, <vscale x 16 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_ta_i32m2:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v10, v12
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 4 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv4i32.nxv8i8.nxv16i8(
+ <vscale x 4 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 16 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 32 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 8 x i32> @intrinsic_vqmaccus_4x8x4_tu_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_tu_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 8 x i32> %a
+}
+
+define <vscale x 8 x i32> @intrinsic_vqmaccus_4x8x4_ta_i32m4(<vscale x 8 x i32> %0, <vscale x 8 x i8> %1, <vscale x 32 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_ta_i32m4:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v12, v16
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 8 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv8i32.nxv8i8.nxv32i8(
+ <vscale x 8 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 32 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32>,
+ <vscale x 8 x i8>,
+ <vscale x 64 x i8>,
+ iXLen, iXLen);
+
+define <vscale x 16 x i32> @intrinsic_vqmaccus_4x8x4_tu_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_tu_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, tu, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 2)
+
+ ret <vscale x 16 x i32> %a
+}
+
+define <vscale x 16 x i32> @intrinsic_vqmaccus_4x8x4_ta_i32m8(<vscale x 16 x i32> %0, <vscale x 8 x i8> %1, <vscale x 64 x i8> %2, iXLen %3) nounwind {
+; CHECK-LABEL: intrinsic_vqmaccus_4x8x4_ta_i32m8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: vl8r.v v24, (a0)
+; CHECK-NEXT: vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT: sf.vqmaccus.4x8x4 v8, v16, v24
+; CHECK-NEXT: ret
+entry:
+ %a = call <vscale x 16 x i32> @llvm.riscv.sf.vqmaccus.4x8x4.nxv16i32.nxv8i8.nxv64i8(
+ <vscale x 16 x i32> %0,
+ <vscale x 8 x i8> %1,
+ <vscale x 64 x i8> %2,
+ iXLen %3, iXLen 3)
+
+ ret <vscale x 16 x i32> %a
+}
diff --git a/llvm/test/MC/RISCV/rvv/xsfvqmacc.s b/llvm/test/MC/RISCV/rvv/xsfvqmacc.s
new file mode 100644
index 000000000000000..ba19f218448699b
--- /dev/null
+++ b/llvm/test/MC/RISCV/rvv/xsfvqmacc.s
@@ -0,0 +1,57 @@
+# RUN: llvm-mc -triple=riscv64 -show-encoding --mattr=+v,+xsfvqmaccqoq,+xsfvqmaccdod %s \
+# RUN: | FileCheck %s --check-prefixes=CHECK-ENCODING,CHECK-INST
+# RUN: not llvm-mc -triple=riscv64 -show-encoding %s 2>&1 \
+# RUN: | FileCheck %s --check-prefix=CHECK-ERROR
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+v,+xsfvqmaccqoq,+xsfvqmaccdod %s \
+# RUN: | llvm-objdump -d --mattr=+v,+xsfvqmaccqoq,+xsfvqmaccdod - \
+# RUN: | FileCheck %s --check-prefix=CHECK-INST
+# RUN: llvm-mc -triple=riscv64 -filetype=obj --mattr=+v,+xsfvqmaccqoq,+xsfvqmaccdod %s \
+# RUN: | llvm-objdump -d - | FileCheck %s --check-prefix=CHECK-UNKNOWN
+
+sf.vqmaccu.2x8x2 v8, v4, v20
+# CHECK-INST: sf.vqmaccu.2x8x2 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xb3]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2))
+# CHECK-UNKNOWN: 5b 24 42 b3 <unknown>
+
+sf.vqmacc.2x8x2 v8, v4, v20
+# CHECK-INST: sf.vqmacc.2x8x2 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xb7]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2))
+# CHECK-UNKNOWN: 5b 24 42 b7 <unknown>
+
+sf.vqmaccus.2x8x2 v8, v4, v20
+# CHECK-INST: sf.vqmaccus.2x8x2 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xbb]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2))
+# CHECK-UNKNOWN: 5b 24 42 bb <unknown>
+
+sf.vqmaccsu.2x8x2 v8, v4, v20
+# CHECK-INST: sf.vqmaccsu.2x8x2 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xbf]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccdod' (SiFive Int8 Matrix Multiplication Instructions (2-by-8 and 8-by-2))
+# CHECK-UNKNOWN: 5b 24 42 bf <unknown>
+
+sf.vqmaccu.4x8x4 v8, v4, v20
+# CHECK-INST: sf.vqmaccu.4x8x4 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xf3]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))
+# CHECK-UNKNOWN: 5b 24 42 f3 <unknown>
+
+sf.vqmacc.4x8x4 v8, v4, v20
+# CHECK-INST: sf.vqmacc.4x8x4 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xf7]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))
+# CHECK-UNKNOWN: 5b 24 42 f7 <unknown>
+
+sf.vqmaccus.4x8x4 v8, v4, v20
+# CHECK-INST: sf.vqmaccus.4x8x4 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xfb]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))
+# CHECK-UNKNOWN: 5b 24 42 fb <unknown>
+
+sf.vqmaccsu.4x8x4 v8, v4, v20
+# CHECK-INST: sf.vqmaccsu.4x8x4 v8, v4, v20
+# CHECK-ENCODING: [0x5b,0x24,0x42,0xff]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))
+# CHECK-UNKNOWN: 5b 24 42 ff <unknown>
diff --git a/llvm/unittests/Support/RISCVISAInfoTest.cpp b/llvm/unittests/Support/RISCVISAInfoTest.cpp
index 90e26a23e87c205..98a19b6fd636502 100644
--- a/llvm/unittests/Support/RISCVISAInfoTest.cpp
+++ b/llvm/unittests/Support/RISCVISAInfoTest.cpp
@@ -713,6 +713,8 @@ R"(All available -march extensions for RISC-V
xcvsimd 1.0
xsfcie 1.0
xsfvcp 1.0
+ xsfvqmaccdod 1.0
+ xsfvqmaccqoq 1.0
xtheadba 1.0
xtheadbb 1.0
xtheadbs 1.0
>From c20affda86b72a51283cbb844bd236b610e38e36 Mon Sep 17 00:00:00 2001
From: Brandon Wu <brandon.wu at sifive.com>
Date: Thu, 5 Oct 2023 03:21:18 -0700
Subject: [PATCH 2/4] fixup! [RISCV] Support Xsfvqmaccdod and Xsfvqmaccqoq
extensions
---
.../Target/RISCV/Disassembler/RISCVDisassembler.cpp | 10 ++++++----
llvm/unittests/Support/RISCVISAInfoTest.cpp | 2 +-
2 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp b/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp
index 845d23945b1d4df..904cc235a8ca986 100644
--- a/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp
+++ b/llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp
@@ -558,10 +558,12 @@ DecodeStatus RISCVDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
"XTHeadVdot custom opcode table");
TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfvcp, DecoderTableXSfvcp32,
"SiFive VCIX custom opcode table");
- TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfvqmaccdod, DecoderTableXSfvqmaccdod32,
- "SiFive Matrix Multiplication Instruction opcode table");
- TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfvqmaccqoq, DecoderTableXSfvqmaccqoq32,
- "SiFive Matrix Multiplication Instruction opcode table");
+ TRY_TO_DECODE_FEATURE(
+ RISCV::FeatureVendorXSfvqmaccdod, DecoderTableXSfvqmaccdod32,
+ "SiFive Matrix Multiplication Instruction opcode table");
+ TRY_TO_DECODE_FEATURE(
+ RISCV::FeatureVendorXSfvqmaccqoq, DecoderTableXSfvqmaccqoq32,
+ "SiFive Matrix Multiplication Instruction opcode table");
TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXSfcie, DecoderTableXSfcie32,
"Sifive CIE custom opcode table");
TRY_TO_DECODE_FEATURE(RISCV::FeatureVendorXCVbitmanip,
diff --git a/llvm/unittests/Support/RISCVISAInfoTest.cpp b/llvm/unittests/Support/RISCVISAInfoTest.cpp
index 98a19b6fd636502..bf353e00df64882 100644
--- a/llvm/unittests/Support/RISCVISAInfoTest.cpp
+++ b/llvm/unittests/Support/RISCVISAInfoTest.cpp
@@ -630,7 +630,7 @@ TEST(getTargetFeatureForExtension, RetrieveTargetFeatureFromOneExt) {
TEST(RiscvExtensionsHelp, CheckExtensions) {
std::string ExpectedOutput =
-R"(All available -march extensions for RISC-V
+ R"(All available -march extensions for RISC-V
Name Version Description
i 2.1 This is a long dummy description
>From a41af664963a4ffefaed3673fc9bb0f323092d87 Mon Sep 17 00:00:00 2001
From: Brandon Wu <brandon.wu at sifive.com>
Date: Fri, 6 Oct 2023 10:22:35 -0700
Subject: [PATCH 3/4] fixup! [RISCV] Support Xsfvqmaccdod and Xsfvqmaccqoq
extensions
---
llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td | 3 ++-
llvm/unittests/Support/RISCVISAInfoTest.cpp | 4 +++-
2 files changed, 5 insertions(+), 2 deletions(-)
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td b/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
index 111358cb0fe080b..22515a6634fd237 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
@@ -347,7 +347,8 @@ multiclass VPseudoVC_XVW<LMULInfo m, DAGOperand RS1Class,
multiclass VPseudoSiFiveVMACC<string mx, VReg vd_type, VReg vs2_type,
string Constraint = ""> {
- def "Pseudo" # NAME # "_VV_" # mx : VPseudoTernaryNoMaskWithPolicy<vd_type, V_M1.vrclass, vs2_type, Constraint>;
+ def "Pseudo" # NAME # "_VV_" # mx
+ : VPseudoTernaryNoMaskWithPolicy<vd_type, V_M1.vrclass, vs2_type, Constraint>;
}
multiclass VPseudoSiFiveVQMACC<string Constraint = ""> {
diff --git a/llvm/unittests/Support/RISCVISAInfoTest.cpp b/llvm/unittests/Support/RISCVISAInfoTest.cpp
index bf353e00df64882..885ec88f8f697d0 100644
--- a/llvm/unittests/Support/RISCVISAInfoTest.cpp
+++ b/llvm/unittests/Support/RISCVISAInfoTest.cpp
@@ -629,8 +629,9 @@ TEST(getTargetFeatureForExtension, RetrieveTargetFeatureFromOneExt) {
}
TEST(RiscvExtensionsHelp, CheckExtensions) {
+ // clang-format off
std::string ExpectedOutput =
- R"(All available -march extensions for RISC-V
+R"(All available -march extensions for RISC-V
Name Version Description
i 2.1 This is a long dummy description
@@ -757,6 +758,7 @@ Experimental extensions
Use -march to specify the target's extension.
For example, clang -march=rv32i_v1p0)";
+ // clang-format on
StringMap<StringRef> DummyMap;
DummyMap["i"] = "This is a long dummy description";
>From b749a6f300c7186d9f8683c7902a2685c84a818b Mon Sep 17 00:00:00 2001
From: Brandon Wu <brandon.wu at sifive.com>
Date: Tue, 10 Oct 2023 20:03:21 -0700
Subject: [PATCH 4/4] fixup! [RISCV] Support Xsfvqmaccdod and Xsfvqmaccqoq
extensions
---
.../include/clang/Basic/riscv_sifive_vector.td | 18 +++++++++---------
.../include/clang/Basic/riscv_vector_common.td | 2 --
.../clang/Support/RISCVVIntrinsicUtils.h | 1 -
clang/lib/Support/RISCVVIntrinsicUtils.cpp | 10 +---------
.../test/Preprocessor/riscv-target-features.c | 18 ++++++++++++++++++
5 files changed, 28 insertions(+), 21 deletions(-)
diff --git a/clang/include/clang/Basic/riscv_sifive_vector.td b/clang/include/clang/Basic/riscv_sifive_vector.td
index 079dc7768fc0580..fb8561a05a0d453 100644
--- a/clang/include/clang/Basic/riscv_sifive_vector.td
+++ b/clang/include/clang/Basic/riscv_sifive_vector.td
@@ -109,21 +109,21 @@ multiclass RVVVQMACCBuiltinSet<list<list<string>> suffixes_prototypes> {
Name = NAME,
HasMasked = false,
Log2LMUL = [0, 1, 2, 3] in
- defm NAME : RVVOutOp1Op2BuiltinSet<NAME, "c", suffixes_prototypes>;
+ defm NAME : RVVOutOp1Op2BuiltinSet<NAME, "i", suffixes_prototypes>;
}
let UnMaskedPolicyScheme = HasPolicyOperand in
let RequiredFeatures = ["Xsfvqmaccdod"] in {
- defm sf_vqmaccu_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44SUvUv"]]>;
- defm sf_vqmacc_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44Svv"]]>;
- defm sf_vqmaccus_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44SUvv"]]>;
- defm sf_vqmaccsu_2x8x2 : RVVVQMACCBuiltinSet<[["", "4", "44SvUv"]]>;
+ defm sf_vqmaccu_2x8x2 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)SUv(FixedSEW:8)Uv"]]>;
+ defm sf_vqmacc_2x8x2 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)Sv(FixedSEW:8)v"]]>;
+ defm sf_vqmaccus_2x8x2 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)SUv(FixedSEW:8)v"]]>;
+ defm sf_vqmaccsu_2x8x2 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)Sv(FixedSEW:8)Uv"]]>;
}
let UnMaskedPolicyScheme = HasPolicyOperand in
let RequiredFeatures = ["Xsfvqmaccqoq"] in {
- defm sf_vqmaccu_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44SUvUv"]]>;
- defm sf_vqmacc_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44Svv"]]>;
- defm sf_vqmaccus_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44SUvv"]]>;
- defm sf_vqmaccsu_4x8x4 : RVVVQMACCBuiltinSet<[["", "4", "44SvUv"]]>;
+ defm sf_vqmaccu_4x8x4 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)SUv(FixedSEW:8)Uv"]]>;
+ defm sf_vqmacc_4x8x4 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)Sv(FixedSEW:8)v"]]>;
+ defm sf_vqmaccus_4x8x4 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)SUv(FixedSEW:8)v"]]>;
+ defm sf_vqmaccsu_4x8x4 : RVVVQMACCBuiltinSet<[["", "v", "vv(FixedSEW:8)Sv(FixedSEW:8)Uv"]]>;
}
diff --git a/clang/include/clang/Basic/riscv_vector_common.td b/clang/include/clang/Basic/riscv_vector_common.td
index b758b7c13b37ec1..141fac9d68e6d54 100644
--- a/clang/include/clang/Basic/riscv_vector_common.td
+++ b/clang/include/clang/Basic/riscv_vector_common.td
@@ -59,8 +59,6 @@
// element type which is four times as wide as the element type of 'v'
// o: computes a vector type identical to what 'v' computes except for the
// element type which is eight times as wide as the element type of 'v'
-// 4: computes a vector type with LMUL identical to what 'v' computes except
-// for the SEW which is 4 times as wide as the SEW of 'v'
// m: computes a vector type identical to what 'v' computes except for the
// element type which is bool
// 0: void type, ignores "t"
diff --git a/clang/include/clang/Support/RISCVVIntrinsicUtils.h b/clang/include/clang/Support/RISCVVIntrinsicUtils.h
index cb76ef598c8ad0c..90d99b7efa8aff9 100644
--- a/clang/include/clang/Support/RISCVVIntrinsicUtils.h
+++ b/clang/include/clang/Support/RISCVVIntrinsicUtils.h
@@ -35,7 +35,6 @@ enum class VectorTypeModifier : uint8_t {
Widening2XVector,
Widening4XVector,
Widening8XVector,
- SEW4XVector,
MaskVector,
Log2EEW3,
Log2EEW4,
diff --git a/clang/lib/Support/RISCVVIntrinsicUtils.cpp b/clang/lib/Support/RISCVVIntrinsicUtils.cpp
index ce46a3f5c662e87..c105db434dc43c9 100644
--- a/clang/lib/Support/RISCVVIntrinsicUtils.cpp
+++ b/clang/lib/Support/RISCVVIntrinsicUtils.cpp
@@ -411,10 +411,6 @@ PrototypeDescriptor::parsePrototypeDescriptor(
PT = BaseTypeModifier::Vector;
VTM = VectorTypeModifier::Widening8XVector;
break;
- case '4':
- PT = BaseTypeModifier::Vector;
- VTM = VectorTypeModifier::SEW4XVector;
- break;
case 'm':
PT = BaseTypeModifier::Vector;
VTM = VectorTypeModifier::MaskVector;
@@ -691,10 +687,6 @@ void RVVType::applyModifier(const PrototypeDescriptor &Transformer) {
LMUL.MulLog2LMUL(3);
Scale = LMUL.getScale(ElementBitwidth);
break;
- case VectorTypeModifier::SEW4XVector:
- ElementBitwidth *= 4;
- Scale = LMUL.getScale(ElementBitwidth);
- break;
case VectorTypeModifier::MaskVector:
ScalarType = ScalarTypeKind::Boolean;
Scale = LMUL.getScale(ElementBitwidth);
@@ -1158,7 +1150,7 @@ void RVVIntrinsic::updateNamesAndPolicy(
SmallVector<PrototypeDescriptor> parsePrototypes(StringRef Prototypes) {
SmallVector<PrototypeDescriptor> PrototypeDescriptors;
- const StringRef Primaries("evwqo4m0ztul");
+ const StringRef Primaries("evwqom0ztul");
while (!Prototypes.empty()) {
size_t Idx = 0;
// Skip over complex prototype because it could contain primitive type
diff --git a/clang/test/Preprocessor/riscv-target-features.c b/clang/test/Preprocessor/riscv-target-features.c
index 242197e3f129a3f..ffdec34ca615fe3 100644
--- a/clang/test/Preprocessor/riscv-target-features.c
+++ b/clang/test/Preprocessor/riscv-target-features.c
@@ -33,6 +33,8 @@
// CHECK-NOT: __riscv_xcvsimd {{.*$}}
// CHECK-NOT: __riscv_xsfcie {{.*$}}
// CHECK-NOT: __riscv_xsfvcp {{.*$}}
+// CHECK-NOT: __riscv_xsfqmaccdod {{.*$}}
+// CHECK-NOT: __riscv_xsfvqmaccqoq {{.*$}}
// CHECK-NOT: __riscv_xtheadba {{.*$}}
// CHECK-NOT: __riscv_xtheadbb {{.*$}}
// CHECK-NOT: __riscv_xtheadbs {{.*$}}
@@ -323,6 +325,22 @@
// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVCP-EXT %s
// CHECK-XSFVCP-EXT: __riscv_xsfvcp 1000000{{$}}
+// RUN: %clang --target=riscv32-unknown-linux-gnu \
+// RUN: -march=rv32ixsfvqmaccdod -x c -E -dM %s \
+// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVQMACCDOD-EXT %s
+// RUN: %clang --target=riscv64-unknown-linux-gnu \
+// RUN: -march=rv64ixsfvqmaccdod -x c -E -dM %s \
+// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVQMACCDOD-EXT %s
+// CHECK-XSFVQMACCDOD-EXT: __riscv_xsfvqmaccdod 1000000{{$}}
+
+// RUN: %clang --target=riscv32-unknown-linux-gnu \
+// RUN: -march=rv32ixsfvqmaccqoq -x c -E -dM %s \
+// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVQMACCQOQ-EXT %s
+// RUN: %clang --target=riscv64-unknown-linux-gnu \
+// RUN: -march=rv64ixsfvqmaccqoq -x c -E -dM %s \
+// RUN: -o - | FileCheck --check-prefix=CHECK-XSFVQMACCQOQ-EXT %s
+// CHECK-XSFVQMACCQOQ-EXT: __riscv_xsfvqmaccqoq 1000000{{$}}
+
// RUN: %clang --target=riscv32-unknown-linux-gnu \
// RUN: -march=rv32ixtheadba -x c -E -dM %s \
// RUN: -o - | FileCheck --check-prefix=CHECK-XTHEADBA-EXT %s
More information about the cfe-commits
mailing list