[llvm] [RISCV] Refine cost on Min/Max reduction (PR #79402)
via llvm-commits
llvm-commits at lists.llvm.org
Wed Jan 24 20:08:10 PST 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-backend-risc-v
Author: Shih-Po Hung (arcbbb)
<details>
<summary>Changes</summary>
This patch is split off from #<!-- -->77342, and follows #<!-- -->79103
- Correct for CodeSize cost that 1 instruction is not included. 3 is from {VMV.S, ReductionOp, VMV.X}
- Add SplitCost which chains a series of VMAX/VMIN/... which scales with LMUL.
- Use MVT to estimate VL.
---
Patch is 107.33 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/79402.diff
5 Files Affected:
- (modified) llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp (+36-7)
- (modified) llvm/test/Analysis/CostModel/RISCV/reduce-max.ll (+70-70)
- (modified) llvm/test/Analysis/CostModel/RISCV/reduce-min.ll (+70-70)
- (modified) llvm/test/Analysis/CostModel/RISCV/reduce-scalable-fp.ll (+42-42)
- (modified) llvm/test/Analysis/CostModel/RISCV/reduce-scalable-int.ll (+48-48)
``````````diff
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 0799af0a7bad550..0ab795ec99d8a87 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -942,13 +942,42 @@ RISCVTTIImpl::getMinMaxReductionCost(Intrinsic::ID IID, VectorType *Ty,
return (LT.first - 1) + 3;
// IR Reduction is composed by two vmv and one rvv reduction instruction.
- InstructionCost BaseCost = 2;
-
- if (CostKind == TTI::TCK_CodeSize)
- return (LT.first - 1) + BaseCost;
-
- unsigned VL = getEstimatedVLFor(Ty);
- return (LT.first - 1) + BaseCost + Log2_32_Ceil(VL);
+ unsigned SplitOp;
+ SmallVector<unsigned, 3> Opcodes;
+ switch (IID) {
+ default:
+ llvm_unreachable("Unsupported intrinsic");
+ case Intrinsic::smax:
+ SplitOp = RISCV::VMAX_VV;
+ Opcodes = {RISCV::VMV_S_X, RISCV::VREDMAX_VS, RISCV::VMV_X_S};
+ break;
+ case Intrinsic::smin:
+ SplitOp = RISCV::VMIN_VV;
+ Opcodes = {RISCV::VMV_S_X, RISCV::VREDMIN_VS, RISCV::VMV_X_S};
+ break;
+ case Intrinsic::umax:
+ SplitOp = RISCV::VMAXU_VV;
+ Opcodes = {RISCV::VMV_S_X, RISCV::VREDMAXU_VS, RISCV::VMV_X_S};
+ break;
+ case Intrinsic::umin:
+ SplitOp = RISCV::VMINU_VV;
+ Opcodes = {RISCV::VMV_S_X, RISCV::VREDMINU_VS, RISCV::VMV_X_S};
+ break;
+ case Intrinsic::maxnum:
+ SplitOp = RISCV::VFMAX_VV;
+ Opcodes = {RISCV::VFMV_S_F, RISCV::VFREDMAX_VS, RISCV::VFMV_F_S};
+ break;
+ case Intrinsic::minnum:
+ SplitOp = RISCV::VFMIN_VV;
+ Opcodes = {RISCV::VFMV_S_F, RISCV::VFREDMIN_VS, RISCV::VFMV_F_S};
+ break;
+ }
+ // Add a cost for data larger than LMUL8
+ InstructionCost SplitCost =
+ (LT.first > 1) ? (LT.first - 1) *
+ getRISCVInstructionCost(SplitOp, LT.second, CostKind)
+ : 0;
+ return SplitCost + getRISCVInstructionCost(Opcodes, LT.second, CostKind);
}
InstructionCost
diff --git a/llvm/test/Analysis/CostModel/RISCV/reduce-max.ll b/llvm/test/Analysis/CostModel/RISCV/reduce-max.ll
index c21b8520112cf88..11fa50d355e833d 100644
--- a/llvm/test/Analysis/CostModel/RISCV/reduce-max.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/reduce-max.ll
@@ -51,14 +51,14 @@ define i32 @reduce_umax_i8(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_umax_i8'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i8 @llvm.vector.reduce.umax.v1i8(<1 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i8 @llvm.vector.reduce.umax.v2i8(<2 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i8 @llvm.vector.reduce.umax.v4i8(<4 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i8 @llvm.vector.reduce.umax.v8i8(<8 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16 = call i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32 = call i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.vector.reduce.umax.v64i8(<64 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V128 = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V1 = call i8 @llvm.vector.reduce.umax.v1i8(<1 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i8 @llvm.vector.reduce.umax.v2i8(<2 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i8 @llvm.vector.reduce.umax.v4i8(<4 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V8 = call i8 @llvm.vector.reduce.umax.v8i8(<8 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V16 = call i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V32 = call i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V64 = call i8 @llvm.vector.reduce.umax.v64i8(<64 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V128 = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> undef)
; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%V1 = call i8 @llvm.vector.reduce.umax.v1i8(<1 x i8> undef)
@@ -85,14 +85,14 @@ define i32 @reduce_umax_i16(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_umax_i16'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i16 @llvm.vector.reduce.umax.v1i16(<1 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i16 @llvm.vector.reduce.umax.v2i16(<2 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i16 @llvm.vector.reduce.umax.v4i16(<4 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V128 = call i16 @llvm.vector.reduce.umax.v128i16(<128 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V1 = call i16 @llvm.vector.reduce.umax.v1i16(<1 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i16 @llvm.vector.reduce.umax.v2i16(<2 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.vector.reduce.umax.v4i16(<4 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V8 = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V128 = call i16 @llvm.vector.reduce.umax.v128i16(<128 x i16> undef)
; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%V1 = call i16 @llvm.vector.reduce.umax.v1i16(<1 x i16> undef)
@@ -115,18 +115,18 @@ define i32 @reduce_umax_i32(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V64 = call i32 @llvm.vector.reduce.umax.v64i32(<64 x i32> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V128 = call i32 @llvm.vector.reduce.umax.v128i32(<128 x i32> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V128 = call i32 @llvm.vector.reduce.umax.v128i32(<128 x i32> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_umax_i32'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i32 @llvm.vector.reduce.umax.v1i32(<1 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i32 @llvm.vector.reduce.umax.v64i32(<64 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V128 = call i32 @llvm.vector.reduce.umax.v128i32(<128 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V1 = call i32 @llvm.vector.reduce.umax.v1i32(<1 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V8 = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V64 = call i32 @llvm.vector.reduce.umax.v64i32(<64 x i32> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i32 @llvm.vector.reduce.umax.v128i32(<128 x i32> undef)
; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%V1 = call i32 @llvm.vector.reduce.umax.v1i32(<1 x i32> undef)
@@ -148,19 +148,19 @@ define i32 @reduce_umax_i64(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V32 = call i64 @llvm.vector.reduce.umax.v32i64(<32 x i64> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V64 = call i64 @llvm.vector.reduce.umax.v64i64(<64 x i64> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V128 = call i64 @llvm.vector.reduce.umax.v128i64(<128 x i64> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i64 @llvm.vector.reduce.umax.v64i64(<64 x i64> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V128 = call i64 @llvm.vector.reduce.umax.v128i64(<128 x i64> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_umax_i64'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32 = call i64 @llvm.vector.reduce.umax.v32i64(<32 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V64 = call i64 @llvm.vector.reduce.umax.v64i64(<64 x i64> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V128 = call i64 @llvm.vector.reduce.umax.v128i64(<128 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V32 = call i64 @llvm.vector.reduce.umax.v32i64(<32 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V64 = call i64 @llvm.vector.reduce.umax.v64i64(<64 x i64> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V128 = call i64 @llvm.vector.reduce.umax.v128i64(<128 x i64> undef)
; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
@@ -221,14 +221,14 @@ define i32 @reduce_smax_i8(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_smax_i8'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i8 @llvm.vector.reduce.smax.v1i8(<1 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V1 = call i8 @llvm.vector.reduce.smax.v1i8(<1 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V8 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%V1 = call i8 @llvm.vector.reduce.smax.v1i8(<1 x i8> undef)
@@ -255,14 +255,14 @@ define i32 @reduce_smax_i16(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_smax_i16'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i16 @llvm.vector.reduce.smax.v1i16(<1 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i16 @llvm.vector.reduce.smax.v2i16(<2 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i16 @llvm.vector.reduce.smax.v4i16(<4 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16 = call i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32 = call i16 @llvm.vector.reduce.smax.v32i16(<32 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V128 = call i16 @llvm.vector.reduce.smax.v128i16(<128 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V1 = call i16 @llvm.vector.reduce.smax.v1i16(<1 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i16 @llvm.vector.reduce.smax.v2i16(<2 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.vector.reduce.smax.v4i16(<4 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V8 = call i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V16 = call i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V32 = call i16 @llvm.vector.reduce.smax.v32i16(<32 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V64 = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> undef)
+; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V128 = call i16 @llvm.vector.reduce.smax.v128i16(<128 x i16> undef)
; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%V1 = call i16 @llvm.vector.reduce.smax.v1i16(<1 x i16> undef)
@@ -285,18 +285,18 @@ define i32 @reduce_smax_i32(i32 %arg) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V32 = call i32 @llvm.vector.reduce.smax.v32i32(<32 x i32> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V64 = call i32 @llvm.vector.reduce.smax.v64i32(<64 x i32> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V128 = call i32 @llvm.vector.reduce.smax.v128i32(<128 x i32> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V128 = call i32 @llvm.vector.reduce.smax.v128i32(<128 x i32> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; SIZE-LABEL: 'reduce_smax_i32'
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i32 @llvm.vector.reduce.smax.v1i32(<1 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2 = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4 = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> undef)
-; SIZE-NEXT: Cost Model: Found an ...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/79402
More information about the llvm-commits
mailing list