[llvm] 5fd5946 - [AArch64][SVE] Allow factors other than 2/4 for load+deinterleave3+store patterns for codegen (#162475)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Oct 16 08:14:05 PDT 2025
Author: Rajveer Singh Bharadwaj
Date: 2025-10-16T20:44:01+05:30
New Revision: 5fd59466f2bdd5338453826d46cc813b949c9ad5
URL: https://github.com/llvm/llvm-project/commit/5fd59466f2bdd5338453826d46cc813b949c9ad5
DIFF: https://github.com/llvm/llvm-project/commit/5fd59466f2bdd5338453826d46cc813b949c9ad5.diff
LOG: [AArch64][SVE] Allow factors other than 2/4 for load+deinterleave3+store patterns for codegen (#162475)
Resolves #159801 and #162068
Added:
Modified:
llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
llvm/test/Transforms/InterleavedAccess/AArch64/scalable-deinterleave-intrinsics.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 91c1f59c1d066..662d84b7a60a8 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -18176,8 +18176,8 @@ bool AArch64TargetLowering::lowerInterleavedStore(Instruction *Store,
bool AArch64TargetLowering::lowerDeinterleaveIntrinsicToLoad(
Instruction *Load, Value *Mask, IntrinsicInst *DI) const {
const unsigned Factor = getDeinterleaveIntrinsicFactor(DI->getIntrinsicID());
- if (Factor != 2 && Factor != 4) {
- LLVM_DEBUG(dbgs() << "Matching ld2 and ld4 patterns failed\n");
+ if (Factor != 2 && Factor != 3 && Factor != 4) {
+ LLVM_DEBUG(dbgs() << "Matching ld2, ld3 and ld4 patterns failed\n");
return false;
}
auto *LI = dyn_cast<LoadInst>(Load);
@@ -18255,8 +18255,8 @@ bool AArch64TargetLowering::lowerInterleaveIntrinsicToStore(
Instruction *Store, Value *Mask,
ArrayRef<Value *> InterleavedValues) const {
unsigned Factor = InterleavedValues.size();
- if (Factor != 2 && Factor != 4) {
- LLVM_DEBUG(dbgs() << "Matching st2 and st4 patterns failed\n");
+ if (Factor != 2 && Factor != 3 && Factor != 4) {
+ LLVM_DEBUG(dbgs() << "Matching st2, st3 and st4 patterns failed\n");
return false;
}
StoreInst *SI = dyn_cast<StoreInst>(Store);
diff --git a/llvm/test/Transforms/InterleavedAccess/AArch64/scalable-deinterleave-intrinsics.ll b/llvm/test/Transforms/InterleavedAccess/AArch64/scalable-deinterleave-intrinsics.ll
index ed9fba3a01965..22ab79d8554cc 100644
--- a/llvm/test/Transforms/InterleavedAccess/AArch64/scalable-deinterleave-intrinsics.ll
+++ b/llvm/test/Transforms/InterleavedAccess/AArch64/scalable-deinterleave-intrinsics.ll
@@ -289,6 +289,225 @@ define void @interleave_wide_nxdouble_factor2(ptr %ptr, <vscale x 4 x double> %l
ret void
}
+define void @deinterleave1_nxi64_factor3(ptr %ptr, <vscale x 4 x i64>* %s1, <vscale x 4 x i64>* %s2, <vscale x 4 x i64>* %s3) #0 {
+; CHECK-LABEL: define void @deinterleave1_nxi64_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], ptr [[S1:%.*]], ptr [[S2:%.*]], ptr [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 0
+; CHECK-NEXT: [[LDN1:%.*]] = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3.sret.nxv2i64(<vscale x 2 x i1> splat (i1 true), ptr [[TMP1]])
+; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN1]], 0
+; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP2]], i64 0)
+; CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN1]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP4]], i64 0)
+; CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN1]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP6]], i64 0)
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 3
+; CHECK-NEXT: [[LDN2:%.*]] = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3.sret.nxv2i64(<vscale x 2 x i1> splat (i1 true), ptr [[TMP8]])
+; CHECK-NEXT: [[TMP9:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN2]], 0
+; CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP3]], <vscale x 2 x i64> [[TMP9]], i64 2)
+; CHECK-NEXT: [[TMP11:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN2]], 1
+; CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP5]], <vscale x 2 x i64> [[TMP11]], i64 2)
+; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN2]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP7]], <vscale x 2 x i64> [[TMP13]], i64 2)
+; CHECK-NEXT: [[TMP15:%.*]] = insertvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } poison, <vscale x 4 x i64> [[TMP10]], 0
+; CHECK-NEXT: [[TMP16:%.*]] = insertvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } [[TMP15]], <vscale x 4 x i64> [[TMP12]], 1
+; CHECK-NEXT: [[TMP17:%.*]] = insertvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } [[TMP16]], <vscale x 4 x i64> [[TMP14]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = extractvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } [[TMP17]], 0
+; CHECK-NEXT: [[TMP19:%.*]] = extractvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } [[TMP17]], 1
+; CHECK-NEXT: [[TMP20:%.*]] = extractvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } [[TMP17]], 2
+; CHECK-NEXT: store <vscale x 4 x i64> [[TMP18]], ptr [[S1]], align 32
+; CHECK-NEXT: store <vscale x 4 x i64> [[TMP19]], ptr [[S2]], align 32
+; CHECK-NEXT: store <vscale x 4 x i64> [[TMP20]], ptr [[S3]], align 32
+; CHECK-NEXT: ret void
+;
+ %wide.vec = load <vscale x 12 x i64>, ptr %ptr, align 8
+ %ldN = tail call { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } @llvm.vector.deinterleave3.nxv12i64(<vscale x 12 x i64> %wide.vec)
+
+ %3 = extractvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } %ldN, 0
+ %4 = extractvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } %ldN, 1
+ %5 = extractvalue { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } %ldN, 2
+
+ store <vscale x 4 x i64> %3, <vscale x 4 x i64>* %s1
+ store <vscale x 4 x i64> %4, <vscale x 4 x i64>* %s2
+ store <vscale x 4 x i64> %5, <vscale x 4 x i64>* %s3
+ ret void
+}
+
+define void @deinterleave2_nxi64_factor3(ptr %ptr, <vscale x 8 x i64>* %s1, <vscale x 8 x i64>* %s2, <vscale x 8 x i64>* %s3) #0 {
+; CHECK-LABEL: define void @deinterleave2_nxi64_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], ptr [[S1:%.*]], ptr [[S2:%.*]], ptr [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 0
+; CHECK-NEXT: [[LDN1:%.*]] = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3.sret.nxv2i64(<vscale x 2 x i1> splat (i1 true), ptr [[TMP1]])
+; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN1]], 0
+; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP2]], i64 0)
+; CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN1]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP4]], i64 0)
+; CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN1]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP6]], i64 0)
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 3
+; CHECK-NEXT: [[LDN2:%.*]] = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3.sret.nxv2i64(<vscale x 2 x i1> splat (i1 true), ptr [[TMP8]])
+; CHECK-NEXT: [[TMP9:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN2]], 0
+; CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP3]], <vscale x 2 x i64> [[TMP9]], i64 2)
+; CHECK-NEXT: [[TMP11:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN2]], 1
+; CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP5]], <vscale x 2 x i64> [[TMP11]], i64 2)
+; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN2]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP7]], <vscale x 2 x i64> [[TMP13]], i64 2)
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 6
+; CHECK-NEXT: [[LDN3:%.*]] = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3.sret.nxv2i64(<vscale x 2 x i1> splat (i1 true), ptr [[TMP15]])
+; CHECK-NEXT: [[TMP16:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN3]], 0
+; CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP10]], <vscale x 2 x i64> [[TMP16]], i64 4)
+; CHECK-NEXT: [[TMP18:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN3]], 1
+; CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP12]], <vscale x 2 x i64> [[TMP18]], i64 4)
+; CHECK-NEXT: [[TMP20:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN3]], 2
+; CHECK-NEXT: [[TMP21:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP14]], <vscale x 2 x i64> [[TMP20]], i64 4)
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 9
+; CHECK-NEXT: [[LDN4:%.*]] = call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld3.sret.nxv2i64(<vscale x 2 x i1> splat (i1 true), ptr [[TMP22]])
+; CHECK-NEXT: [[TMP23:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN4]], 0
+; CHECK-NEXT: [[TMP24:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP17]], <vscale x 2 x i64> [[TMP23]], i64 6)
+; CHECK-NEXT: [[TMP25:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN4]], 1
+; CHECK-NEXT: [[TMP26:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP19]], <vscale x 2 x i64> [[TMP25]], i64 6)
+; CHECK-NEXT: [[TMP27:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[LDN4]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP21]], <vscale x 2 x i64> [[TMP27]], i64 6)
+; CHECK-NEXT: [[TMP29:%.*]] = insertvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } poison, <vscale x 8 x i64> [[TMP24]], 0
+; CHECK-NEXT: [[TMP30:%.*]] = insertvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } [[TMP29]], <vscale x 8 x i64> [[TMP26]], 1
+; CHECK-NEXT: [[TMP31:%.*]] = insertvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } [[TMP30]], <vscale x 8 x i64> [[TMP28]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = extractvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } [[TMP31]], 0
+; CHECK-NEXT: [[TMP33:%.*]] = extractvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } [[TMP31]], 1
+; CHECK-NEXT: [[TMP34:%.*]] = extractvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } [[TMP31]], 2
+; CHECK-NEXT: store <vscale x 8 x i64> [[TMP32]], ptr [[S1]], align 64
+; CHECK-NEXT: store <vscale x 8 x i64> [[TMP33]], ptr [[S2]], align 64
+; CHECK-NEXT: store <vscale x 8 x i64> [[TMP34]], ptr [[S3]], align 64
+; CHECK-NEXT: ret void
+;
+ %wide.vec = load <vscale x 24 x i64>, ptr %ptr, align 8
+ %ldN = tail call { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } @llvm.vector.deinterleave3.nxv24i64(<vscale x 24 x i64> %wide.vec)
+
+ %3 = extractvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } %ldN, 0
+ %4 = extractvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } %ldN, 1
+ %5 = extractvalue { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } %ldN, 2
+
+ store <vscale x 8 x i64> %3, <vscale x 8 x i64>* %s1
+ store <vscale x 8 x i64> %4, <vscale x 8 x i64>* %s2
+ store <vscale x 8 x i64> %5, <vscale x 8 x i64>* %s3
+ ret void
+}
+
+define void @deinterleave_neg1_nxi64_factor3(ptr %ptr, <vscale x 1 x i64>* %s1, <vscale x 1 x i64>* %s2, <vscale x 1 x i64>* %s3) #0 {
+; CHECK-LABEL: define void @deinterleave_neg1_nxi64_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], ptr [[S1:%.*]], ptr [[S2:%.*]], ptr [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 3 x i64>, ptr [[PTR]], align 8
+; CHECK-NEXT: [[LDN:%.*]] = tail call { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } @llvm.vector.deinterleave3.nxv3i64(<vscale x 3 x i64> [[WIDE_VEC]])
+; CHECK-NEXT: [[TMP1:%.*]] = extractvalue { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } [[LDN]], 0
+; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } [[LDN]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } [[LDN]], 2
+; CHECK-NEXT: store <vscale x 1 x i64> [[TMP1]], ptr [[S1]], align 8
+; CHECK-NEXT: store <vscale x 1 x i64> [[TMP2]], ptr [[S2]], align 8
+; CHECK-NEXT: store <vscale x 1 x i64> [[TMP3]], ptr [[S3]], align 8
+; CHECK-NEXT: ret void
+;
+ %wide.vec = load <vscale x 3 x i64>, ptr %ptr, align 8
+ %ldN = tail call { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } @llvm.vector.deinterleave3.nxv3i64(<vscale x 3 x i64> %wide.vec)
+
+ %3 = extractvalue { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } %ldN, 0
+ %4 = extractvalue { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } %ldN, 1
+ %5 = extractvalue { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } %ldN, 2
+
+ store <vscale x 1 x i64> %3, <vscale x 1 x i64>* %s1
+ store <vscale x 1 x i64> %4, <vscale x 1 x i64>* %s2
+ store <vscale x 1 x i64> %5, <vscale x 1 x i64>* %s3
+ ret void
+}
+
+define void @deinterleave_neg2_nxi8_factor3(ptr %ptr, <vscale x 8 x i8>* %s1, <vscale x 8 x i8>* %s2, <vscale x 8 x i8>* %s3) #0 {
+; CHECK-LABEL: define void @deinterleave_neg2_nxi8_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], ptr [[S1:%.*]], ptr [[S2:%.*]], ptr [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 24 x i8>, ptr [[PTR]], align 8
+; CHECK-NEXT: [[LDN:%.*]] = tail call { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } @llvm.vector.deinterleave3.nxv24i8(<vscale x 24 x i8> [[WIDE_VEC]])
+; CHECK-NEXT: [[TMP1:%.*]] = extractvalue { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } [[LDN]], 0
+; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } [[LDN]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } [[LDN]], 2
+; CHECK-NEXT: store <vscale x 8 x i8> [[TMP1]], ptr [[S1]], align 8
+; CHECK-NEXT: store <vscale x 8 x i8> [[TMP2]], ptr [[S2]], align 8
+; CHECK-NEXT: store <vscale x 8 x i8> [[TMP3]], ptr [[S3]], align 8
+; CHECK-NEXT: ret void
+;
+ %wide.vec = load <vscale x 24 x i8>, ptr %ptr, align 8
+ %ldN = tail call { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } @llvm.vector.deinterleave3.nxv12i8(<vscale x 24 x i8> %wide.vec)
+
+ %3 = extractvalue { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } %ldN, 0
+ %4 = extractvalue { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } %ldN, 1
+ %5 = extractvalue { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } %ldN, 2
+
+ store <vscale x 8 x i8> %3, <vscale x 8 x i8>* %s1
+ store <vscale x 8 x i8> %4, <vscale x 8 x i8>* %s2
+ store <vscale x 8 x i8> %5, <vscale x 8 x i8>* %s3
+ ret void
+}
+
+define void @interleave1_nxi64_factor3(ptr %ptr, <vscale x 8 x i64> %s1, <vscale x 8 x i64> %s2, <vscale x 8 x i64> %s3) #0 {
+; CHECK-LABEL: define void @interleave1_nxi64_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], <vscale x 8 x i64> [[S1:%.*]], <vscale x 8 x i64> [[S2:%.*]], <vscale x 8 x i64> [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S1]], i64 0)
+; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S2]], i64 0)
+; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S3]], i64 0)
+; CHECK-NEXT: call void @llvm.aarch64.sve.st3.nxv2i64(<vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], <vscale x 2 x i64> [[TMP4]], <vscale x 2 x i1> splat (i1 true), ptr [[TMP1]])
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 3
+; CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S1]], i64 2)
+; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S2]], i64 2)
+; CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S3]], i64 2)
+; CHECK-NEXT: call void @llvm.aarch64.sve.st3.nxv2i64(<vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], <vscale x 2 x i64> [[TMP8]], <vscale x 2 x i1> splat (i1 true), ptr [[TMP5]])
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 6
+; CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S1]], i64 4)
+; CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S2]], i64 4)
+; CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S3]], i64 4)
+; CHECK-NEXT: call void @llvm.aarch64.sve.st3.nxv2i64(<vscale x 2 x i64> [[TMP10]], <vscale x 2 x i64> [[TMP11]], <vscale x 2 x i64> [[TMP12]], <vscale x 2 x i1> splat (i1 true), ptr [[TMP9]])
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 9
+; CHECK-NEXT: [[TMP14:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S1]], i64 6)
+; CHECK-NEXT: [[TMP15:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S2]], i64 6)
+; CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[S3]], i64 6)
+; CHECK-NEXT: call void @llvm.aarch64.sve.st3.nxv2i64(<vscale x 2 x i64> [[TMP14]], <vscale x 2 x i64> [[TMP15]], <vscale x 2 x i64> [[TMP16]], <vscale x 2 x i1> splat (i1 true), ptr [[TMP13]])
+; CHECK-NEXT: ret void
+;
+ %interleave = tail call <vscale x 24 x i64> @llvm.vector.interleave3.nxv24i64(<vscale x 8 x i64> %s1, <vscale x 8 x i64> %s2, <vscale x 8 x i64> %s3)
+
+ store <vscale x 24 x i64> %interleave, ptr %ptr, align 4
+ ret void
+}
+
+define void @interleave2_nxi64_factor3(ptr %ptr, <vscale x 4 x i64> %s1, <vscale x 4 x i64> %s2, <vscale x 4 x i64> %s3) #0 {
+; CHECK-LABEL: define void @interleave2_nxi64_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], <vscale x 4 x i64> [[S1:%.*]], <vscale x 4 x i64> [[S2:%.*]], <vscale x 4 x i64> [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[S1]], i64 0)
+; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[S2]], i64 0)
+; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[S3]], i64 0)
+; CHECK-NEXT: call void @llvm.aarch64.sve.st3.nxv2i64(<vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], <vscale x 2 x i64> [[TMP4]], <vscale x 2 x i1> splat (i1 true), ptr [[TMP1]])
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr <vscale x 2 x i64>, ptr [[PTR]], i64 3
+; CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[S1]], i64 2)
+; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[S2]], i64 2)
+; CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[S3]], i64 2)
+; CHECK-NEXT: call void @llvm.aarch64.sve.st3.nxv2i64(<vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], <vscale x 2 x i64> [[TMP8]], <vscale x 2 x i1> splat (i1 true), ptr [[TMP5]])
+; CHECK-NEXT: ret void
+;
+ %interleave = tail call <vscale x 12 x i64> @llvm.vector.interleave3.nxv12i64(<vscale x 4 x i64> %s1, <vscale x 4 x i64> %s2, <vscale x 4 x i64> %s3)
+
+ store <vscale x 12 x i64> %interleave, ptr %ptr, align 4
+ ret void
+}
+
+define void @interleave_neg_nxi8_factor3(ptr %ptr, <vscale x 8 x i8> %s1, <vscale x 8 x i8> %s2, <vscale x 8 x i8> %s3) #0 {
+; CHECK-LABEL: define void @interleave_neg_nxi8_factor3
+; CHECK-SAME: (ptr [[PTR:%.*]], <vscale x 8 x i8> [[S1:%.*]], <vscale x 8 x i8> [[S2:%.*]], <vscale x 8 x i8> [[S3:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[INTERLEAVE:%.*]] = tail call <vscale x 24 x i8> @llvm.vector.interleave3.nxv24i8(<vscale x 8 x i8> [[S1]], <vscale x 8 x i8> [[S2]], <vscale x 8 x i8> [[S3]])
+; CHECK-NEXT: store <vscale x 24 x i8> [[INTERLEAVE]], ptr [[PTR]], align 4
+; CHECK-NEXT: ret void
+;
+ %interleave = tail call <vscale x 24 x i8> @llvm.vector.interleave3.nxv24i8(<vscale x 8 x i8> %s1, <vscale x 8 x i8> %s2, <vscale x 8 x i8> %s3)
+
+ store <vscale x 24 x i8> %interleave, ptr %ptr, align 4
+ ret void
+}
+
declare { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.vector.deinterleave2.nxv32i8(<vscale x 32 x i8>)
declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.vector.deinterleave2.nxv16i16(<vscale x 16 x i16>)
declare { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave2.nxv8i32(<vscale x 8 x i32>)
@@ -312,4 +531,15 @@ declare <vscale x 4 x ptr> @llvm.vector.interleave2.nxv4p0(<vscale x 2 x ptr>, <
; Larger interleaves to test 'legalization'
declare <vscale x 8 x double> @llvm.vector.interleave2.nxv8f64(<vscale x 4 x double>, <vscale x 4 x double>)
+; De-Interleaves with Factor=3
+declare { <vscale x 1 x i64>, <vscale x 1 x i64>, <vscale x 1 x i64> } @llvm.vector.deinterleave3.nxv3i64(<vscale x 3 x i64>)
+declare { <vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64> } @llvm.vector.deinterleave3.nxv12i64(<vscale x 12 x i64>)
+declare { <vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64> } @llvm.vector.deinterleave3.nxv24i64(<vscale x 24 x i64>)
+declare { <vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8> } @llvm.vector.deinterleave3.nxv24i8(<vscale x 24 x i8>)
+
+; Interleaves with Factor=3
+declare <vscale x 24 x i8> @llvm.vector.interleave3.nxv24i8(<vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i8>)
+declare <vscale x 24 x i64> @llvm.vector.interleave3.nxv24i64(<vscale x 8 x i64>, <vscale x 8 x i64>, <vscale x 8 x i64>)
+declare <vscale x 12 x i64> @llvm.vector.interleave3.nxv12i64(<vscale x 4 x i64>, <vscale x 4 x i64>, <vscale x 4 x i64>)
+
attributes #0 = { vscale_range(1,16) "target-features"="+sve" }
More information about the llvm-commits
mailing list