diff options
Diffstat (limited to 'llvm/test/Analysis/CostModel/ARM/mve-vecreduce-add.ll')
| -rw-r--r-- | llvm/test/Analysis/CostModel/ARM/mve-vecreduce-add.ll | 858 |
1 files changed, 429 insertions, 429 deletions
diff --git a/llvm/test/Analysis/CostModel/ARM/mve-vecreduce-add.ll b/llvm/test/Analysis/CostModel/ARM/mve-vecreduce-add.ll index 8edc2e6fc156..5f1bce902b93 100644 --- a/llvm/test/Analysis/CostModel/ARM/mve-vecreduce-add.ll +++ b/llvm/test/Analysis/CostModel/ARM/mve-vecreduce-add.ll @@ -1,16 +1,16 @@ ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py -; RUN: opt < %s -S -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s +; RUN: opt -passes="print<cost-model>" -cost-kind=all 2>&1 -disable-output -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve < %s | FileCheck %s target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64" define void @add_i8() { ; CHECK-LABEL: 'add_i8' -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0 = call i8 @llvm.vector.reduce.add.v1i8(<1 x i8> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1 = call i8 @llvm.vector.reduce.add.v2i8(<2 x i8> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2 = call i8 @llvm.vector.reduce.add.v4i8(<4 x i8> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3 = call i8 @llvm.vector.reduce.add.v8i8(<8 x i8> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a4 = call i8 @llvm.vector.reduce.add.v16i8(<16 x i8> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0 = call i8 @llvm.vector.reduce.add.v1i8(<1 x i8> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1 = call i8 @llvm.vector.reduce.add.v2i8(<2 x i8> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2 = call i8 @llvm.vector.reduce.add.v4i8(<4 x i8> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3 = call i8 @llvm.vector.reduce.add.v8i8(<8 x i8> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a4 = call i8 @llvm.vector.reduce.add.v16i8(<16 x i8> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0 = call i8 @llvm.vector.reduce.add.v1i8(<1 x i8> undef) @@ -27,32 +27,32 @@ define void @add_i8() { define void @add_i16() { ; CHECK-LABEL: 'add_i16' -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0za = zext <1 x i8> undef to <1 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0z = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sa = sext <1 x i8> undef to <1 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0s = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a1za = zext <2 x i8> undef to <2 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1z = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1sa = sext <2 x i8> undef to <2 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1s = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2za = zext <4 x i8> undef to <4 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2z = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2sa = sext <4 x i8> undef to <4 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2s = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3za = zext <8 x i8> undef to <8 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3z = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3sa = sext <8 x i8> undef to <8 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3s = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a4za = zext <16 x i8> undef to <16 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a4z = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a4sa = sext <16 x i8> undef to <16 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a4s = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5 = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a6 = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a8 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a9 = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0za = zext <1 x i8> undef to <1 x i16> +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0z = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0za) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0sa = sext <1 x i8> undef to <1 x i16> +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0s = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a1za = zext <2 x i8> undef to <2 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1z = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1za) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a1sa = sext <2 x i8> undef to <2 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1s = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a2za = zext <4 x i8> undef to <4 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2z = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2za) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a2sa = sext <4 x i8> undef to <4 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2s = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3za = zext <8 x i8> undef to <8 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3z = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3za) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3sa = sext <8 x i8> undef to <8 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3s = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a4za = zext <16 x i8> undef to <16 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a4z = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4za) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a4sa = sext <16 x i8> undef to <16 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a4s = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4sa) +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a5 = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a6 = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a8 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a9 = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0za = zext <1 x i8> undef to <1 x i16> %a0z = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0za) @@ -99,52 +99,52 @@ define void @add_i16() { define void @add_i32() { ; CHECK-LABEL: 'add_i32' -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0za = zext <1 x i8> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sa = sext <1 x i8> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a1za = zext <2 x i8> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1sa = sext <2 x i8> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a2za = zext <4 x i8> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a2sa = sext <4 x i8> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a3za = zext <8 x i8> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a3z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a3sa = sext <8 x i8> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a3s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a4za = zext <16 x i8> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a4z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a4sa = sext <16 x i8> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a4s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5za = zext <1 x i16> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5sa = sext <1 x i16> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a6za = zext <2 x i16> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a6z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a6sa = sext <2 x i16> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a6s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7za = zext <4 x i16> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7sa = sext <4 x i16> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a8za = zext <8 x i16> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a8z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a8sa = sext <8 x i16> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a8s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a9za = zext <16 x i16> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a9z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a9sa = sext <16 x i16> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a9s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a10 = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a11 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a12 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a13 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a14 = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0za = zext <1 x i8> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0za) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0sa = sext <1 x i8> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a1za = zext <2 x i8> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1za) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a1sa = sext <2 x i8> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a2za = zext <4 x i8> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2za) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a2sa = sext <4 x i8> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a3za = zext <8 x i8> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a3z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3za) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a3sa = sext <8 x i8> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a3s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a4za = zext <16 x i8> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a4z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4za) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a4sa = sext <16 x i8> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a4s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a5za = zext <1 x i16> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a5z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5za) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a5sa = sext <1 x i16> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a5s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a6za = zext <2 x i16> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a6z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6za) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a6sa = sext <2 x i16> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a6s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7za = zext <4 x i16> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7za) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7sa = sext <4 x i16> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a8za = zext <8 x i16> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a8z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8za) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a8sa = sext <8 x i16> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a8s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:40 CodeSize:1 Lat:1 SizeLat:1 for: %a9za = zext <16 x i16> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a9z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9za) +; CHECK-NEXT: Cost Model: Found costs of RThru:40 CodeSize:1 Lat:1 SizeLat:1 for: %a9sa = sext <16 x i16> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a9s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9sa) +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a10 = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a11 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a12 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a13 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a14 = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0za = zext <1 x i8> undef to <1 x i32> %a0z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0za) @@ -221,72 +221,72 @@ define void @add_i32() { define void @add_i64() { ; CHECK-LABEL: 'add_i64' -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a0za = zext <1 x i8> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a0z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a0sa = sext <1 x i8> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a0s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1za = zext <2 x i8> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a1z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a1sa = sext <2 x i8> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a1s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a2za = zext <4 x i8> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a2z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %a2sa = sext <4 x i8> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a2s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a3za = zext <8 x i8> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a3z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 330 for instruction: %a3sa = sext <8 x i8> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a3s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 298 for instruction: %a4za = zext <16 x i8> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a4z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 1322 for instruction: %a4sa = sext <16 x i8> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a4s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a5za = zext <1 x i16> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a5z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %a5sa = sext <1 x i16> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a5s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a6za = zext <2 x i16> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a6z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a6sa = sext <2 x i16> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a6s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a7za = zext <4 x i16> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a7z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %a7sa = sext <4 x i16> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a7s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a8za = zext <8 x i16> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a8z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 330 for instruction: %a8sa = sext <8 x i16> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a8s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 296 for instruction: %a9za = zext <16 x i16> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a9z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 1320 for instruction: %a9sa = sext <16 x i16> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a9s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a10za = zext <1 x i32> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a10z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a10sa = sext <1 x i32> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a10s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a11za = zext <2 x i32> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a11z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %a11sa = sext <2 x i32> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a11s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a12za = zext <4 x i32> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a12z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %a12sa = sext <4 x i32> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a12s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %a13za = zext <8 x i32> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a13z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 264 for instruction: %a13sa = sext <8 x i32> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a13s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 288 for instruction: %a14za = zext <16 x i32> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a14z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14za) -; CHECK-NEXT: Cost Model: Found an estimated cost of 1056 for instruction: %a14sa = sext <16 x i32> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a14s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14sa) -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a15 = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a16 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a17 = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a18 = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a19 = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> undef) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a0za = zext <1 x i8> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a0z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0za) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a0sa = sext <1 x i8> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a0s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a1za = zext <2 x i8> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a1z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1za) +; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:10 Lat:20 SizeLat:20 for: %a1sa = sext <2 x i8> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a1s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a2za = zext <4 x i8> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a2z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2za) +; CHECK-NEXT: Cost Model: Found costs of RThru:82 CodeSize:1 Lat:1 SizeLat:1 for: %a2sa = sext <4 x i8> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a2s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a3za = zext <8 x i8> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a3z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3za) +; CHECK-NEXT: Cost Model: Found costs of RThru:330 CodeSize:1 Lat:1 SizeLat:1 for: %a3sa = sext <8 x i8> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a3s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:298 CodeSize:1 Lat:1 SizeLat:1 for: %a4za = zext <16 x i8> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a4z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4za) +; CHECK-NEXT: Cost Model: Found costs of RThru:1322 CodeSize:1 Lat:1 SizeLat:1 for: %a4sa = sext <16 x i8> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a4s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a5za = zext <1 x i16> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a5z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5za) +; CHECK-NEXT: Cost Model: Found costs of RThru:36 CodeSize:1 Lat:1 SizeLat:1 for: %a5sa = sext <1 x i16> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a5s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a6za = zext <2 x i16> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a6z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6za) +; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:10 Lat:20 SizeLat:20 for: %a6sa = sext <2 x i16> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a6s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a7za = zext <4 x i16> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a7z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7za) +; CHECK-NEXT: Cost Model: Found costs of RThru:82 CodeSize:1 Lat:1 SizeLat:1 for: %a7sa = sext <4 x i16> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a7s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a8za = zext <8 x i16> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a8z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8za) +; CHECK-NEXT: Cost Model: Found costs of RThru:330 CodeSize:1 Lat:1 SizeLat:1 for: %a8sa = sext <8 x i16> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a8s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:296 CodeSize:1 Lat:1 SizeLat:1 for: %a9za = zext <16 x i16> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a9z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9za) +; CHECK-NEXT: Cost Model: Found costs of RThru:1320 CodeSize:1 Lat:1 SizeLat:1 for: %a9sa = sext <16 x i16> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a9s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a10za = zext <1 x i32> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a10z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10za) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a10sa = sext <1 x i32> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a10s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a11za = zext <2 x i32> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a11z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11za) +; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %a11sa = sext <2 x i32> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a11s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a12za = zext <4 x i32> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a12z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12za) +; CHECK-NEXT: Cost Model: Found costs of RThru:66 CodeSize:1 Lat:1 SizeLat:1 for: %a12sa = sext <4 x i32> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a12s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:72 CodeSize:1 Lat:1 SizeLat:1 for: %a13za = zext <8 x i32> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a13z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13za) +; CHECK-NEXT: Cost Model: Found costs of RThru:264 CodeSize:1 Lat:1 SizeLat:1 for: %a13sa = sext <8 x i32> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a13s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13sa) +; CHECK-NEXT: Cost Model: Found costs of RThru:288 CodeSize:1 Lat:1 SizeLat:1 for: %a14za = zext <16 x i32> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a14z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14za) +; CHECK-NEXT: Cost Model: Found costs of RThru:1056 CodeSize:1 Lat:1 SizeLat:1 for: %a14sa = sext <16 x i32> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a14s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14sa) +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a15 = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a16 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a17 = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a18 = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a19 = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> undef) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0za = zext <1 x i8> undef to <1 x i64> %a0z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0za) @@ -393,17 +393,17 @@ define void @add_i64() { define void @mla_i8() { ; CHECK-LABEL: 'mla_i8' -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0m = mul <1 x i8> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0 = call i8 @llvm.vector.reduce.add.v1i8(<1 x i8> %a0m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a1m = mul <2 x i8> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1 = call i8 @llvm.vector.reduce.add.v2i8(<2 x i8> %a1m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2m = mul <4 x i8> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2 = call i8 @llvm.vector.reduce.add.v4i8(<4 x i8> %a2m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3m = mul <8 x i8> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3 = call i8 @llvm.vector.reduce.add.v8i8(<8 x i8> %a3m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a4m = mul <16 x i8> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a4 = call i8 @llvm.vector.reduce.add.v16i8(<16 x i8> %a4m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a0m = mul <1 x i8> undef, undef +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0 = call i8 @llvm.vector.reduce.add.v1i8(<1 x i8> %a0m) +; CHECK-NEXT: Cost Model: Found costs of 10 for: %a1m = mul <2 x i8> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1 = call i8 @llvm.vector.reduce.add.v2i8(<2 x i8> %a1m) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2m = mul <4 x i8> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2 = call i8 @llvm.vector.reduce.add.v4i8(<4 x i8> %a2m) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3m = mul <8 x i8> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3 = call i8 @llvm.vector.reduce.add.v8i8(<8 x i8> %a3m) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a4m = mul <16 x i8> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a4 = call i8 @llvm.vector.reduce.add.v16i8(<16 x i8> %a4m) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0m = mul <1 x i8> undef, undef %a0 = call i8 @llvm.vector.reduce.add.v1i8(<1 x i8> %a0m) @@ -425,57 +425,57 @@ define void @mla_i8() { define void @mla_i16() { ; CHECK-LABEL: 'mla_i16' -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0za = zext <1 x i8> undef to <1 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0zb = zext <1 x i8> undef to <1 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0zm = mul <1 x i16> %a0za, %a0zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0z = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sa = sext <1 x i8> undef to <1 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sb = sext <1 x i8> undef to <1 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sm = mul <1 x i16> %a0sa, %a0sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0s = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a1za = zext <2 x i8> undef to <2 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a1zb = zext <2 x i8> undef to <2 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %a1zm = mul <2 x i16> %a1za, %a1zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1z = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1sa = sext <2 x i8> undef to <2 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1sb = sext <2 x i8> undef to <2 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %a1sm = mul <2 x i16> %a1sa, %a1sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1s = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2za = zext <4 x i8> undef to <4 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2zb = zext <4 x i8> undef to <4 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2zm = mul <4 x i16> %a2za, %a2zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2z = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2sa = sext <4 x i8> undef to <4 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2sb = sext <4 x i8> undef to <4 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2sm = mul <4 x i16> %a2sa, %a2sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2s = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3za = zext <8 x i8> undef to <8 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3zb = zext <8 x i8> undef to <8 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3zm = mul <8 x i16> %a3za, %a3zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3z = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3sa = sext <8 x i8> undef to <8 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3sb = sext <8 x i8> undef to <8 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3sm = mul <8 x i16> %a3sa, %a3sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a3s = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a4za = zext <16 x i8> undef to <16 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a4zb = zext <16 x i8> undef to <16 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a4zm = mul <16 x i16> %a4za, %a4zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a4z = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a4sa = sext <16 x i8> undef to <16 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a4sb = sext <16 x i8> undef to <16 x i16> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a4sm = mul <16 x i16> %a4sa, %a4sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a4s = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5m = mul <1 x i16> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5 = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a5m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a6m = mul <2 x i16> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a6 = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a6m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7m = mul <4 x i16> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a7m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a8m = mul <8 x i16> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a8 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a8m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a9m = mul <16 x i16> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a9 = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a9m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0za = zext <1 x i8> undef to <1 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0zb = zext <1 x i8> undef to <1 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a0zm = mul <1 x i16> %a0za, %a0zb +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0z = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0sa = sext <1 x i8> undef to <1 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0sb = sext <1 x i8> undef to <1 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a0sm = mul <1 x i16> %a0sa, %a0sb +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0s = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a0sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a1za = zext <2 x i8> undef to <2 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a1zb = zext <2 x i8> undef to <2 x i16> +; CHECK-NEXT: Cost Model: Found costs of 26 for: %a1zm = mul <2 x i16> %a1za, %a1zb +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1z = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a1sa = sext <2 x i8> undef to <2 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a1sb = sext <2 x i8> undef to <2 x i16> +; CHECK-NEXT: Cost Model: Found costs of 26 for: %a1sm = mul <2 x i16> %a1sa, %a1sb +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1s = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a1sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a2za = zext <4 x i8> undef to <4 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a2zb = zext <4 x i8> undef to <4 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2zm = mul <4 x i16> %a2za, %a2zb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2z = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a2sa = sext <4 x i8> undef to <4 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a2sb = sext <4 x i8> undef to <4 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2sm = mul <4 x i16> %a2sa, %a2sb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2s = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a2sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3za = zext <8 x i8> undef to <8 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3zb = zext <8 x i8> undef to <8 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3zm = mul <8 x i16> %a3za, %a3zb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3z = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3sa = sext <8 x i8> undef to <8 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3sb = sext <8 x i8> undef to <8 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3sm = mul <8 x i16> %a3sa, %a3sb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a3s = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a3sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a4za = zext <16 x i8> undef to <16 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a4zb = zext <16 x i8> undef to <16 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a4zm = mul <16 x i16> %a4za, %a4zb +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a4z = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a4sa = sext <16 x i8> undef to <16 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a4sb = sext <16 x i8> undef to <16 x i16> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a4sm = mul <16 x i16> %a4sa, %a4sb +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a4s = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a4sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a5m = mul <1 x i16> undef, undef +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a5 = call i16 @llvm.vector.reduce.add.v1i16(<1 x i16> %a5m) +; CHECK-NEXT: Cost Model: Found costs of 10 for: %a6m = mul <2 x i16> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a6 = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %a6m) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7m = mul <4 x i16> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7 = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %a7m) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a8m = mul <8 x i16> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a8 = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %a8m) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a9m = mul <16 x i16> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a9 = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %a9m) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0za = zext <1 x i8> undef to <1 x i16> %a0zb = zext <1 x i8> undef to <1 x i16> @@ -547,97 +547,97 @@ define void @mla_i16() { define void @mla_i32() { ; CHECK-LABEL: 'mla_i32' -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0za = zext <1 x i8> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0zb = zext <1 x i8> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0zm = mul <1 x i32> %a0za, %a0zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sa = sext <1 x i8> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sb = sext <1 x i8> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a0sm = mul <1 x i32> %a0sa, %a0sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a1za = zext <2 x i8> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a1zb = zext <2 x i8> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %a1zm = mul <2 x i32> %a1za, %a1zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1sa = sext <2 x i8> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1sb = sext <2 x i8> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %a1sm = mul <2 x i32> %a1sa, %a1sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a1s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a2za = zext <4 x i8> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a2zb = zext <4 x i8> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2zm = mul <4 x i32> %a2za, %a2zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a2sa = sext <4 x i8> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a2sb = sext <4 x i8> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2sm = mul <4 x i32> %a2sa, %a2sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a2s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a3za = zext <8 x i8> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a3zb = zext <8 x i8> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a3zm = mul <8 x i32> %a3za, %a3zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a3z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a3sa = sext <8 x i8> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a3sb = sext <8 x i8> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a3sm = mul <8 x i32> %a3sa, %a3sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a3s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a4za = zext <16 x i8> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a4zb = zext <16 x i8> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a4zm = mul <16 x i32> %a4za, %a4zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a4z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a4sa = sext <16 x i8> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a4sb = sext <16 x i8> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a4sm = mul <16 x i32> %a4sa, %a4sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a4s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5za = zext <1 x i16> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5zb = zext <1 x i16> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5zm = mul <1 x i32> %a5za, %a5zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5sa = sext <1 x i16> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5sb = sext <1 x i16> undef to <1 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a5sm = mul <1 x i32> %a5sa, %a5sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a6za = zext <2 x i16> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a6zb = zext <2 x i16> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %a6zm = mul <2 x i32> %a6za, %a6zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a6z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a6sa = sext <2 x i16> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a6sb = sext <2 x i16> undef to <2 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %a6sm = mul <2 x i32> %a6sa, %a6sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a6s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7za = zext <4 x i16> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7zb = zext <4 x i16> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7zm = mul <4 x i32> %a7za, %a7zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7sa = sext <4 x i16> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7sb = sext <4 x i16> undef to <4 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7sm = mul <4 x i32> %a7sa, %a7sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a7s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a8za = zext <8 x i16> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a8zb = zext <8 x i16> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a8zm = mul <8 x i32> %a8za, %a8zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a8z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a8sa = sext <8 x i16> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a8sb = sext <8 x i16> undef to <8 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a8sm = mul <8 x i32> %a8sa, %a8sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a8s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a9za = zext <16 x i16> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a9zb = zext <16 x i16> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a9zm = mul <16 x i32> %a9za, %a9zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a9z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a9sa = sext <16 x i16> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a9sb = sext <16 x i16> undef to <16 x i32> -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a9sm = mul <16 x i32> %a9sa, %a9sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a9s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a10m = mul <1 x i32> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a10 = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a10m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %a11m = mul <2 x i32> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %a11 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a11m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a12m = mul <4 x i32> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %a12 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a12m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a13m = mul <8 x i32> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a13 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a13m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a14m = mul <16 x i32> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a14 = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a14m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0za = zext <1 x i8> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0zb = zext <1 x i8> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a0zm = mul <1 x i32> %a0za, %a0zb +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0sa = sext <1 x i8> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a0sb = sext <1 x i8> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a0sm = mul <1 x i32> %a0sa, %a0sb +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a0s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a0sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a1za = zext <2 x i8> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a1zb = zext <2 x i8> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of 26 for: %a1zm = mul <2 x i32> %a1za, %a1zb +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a1sa = sext <2 x i8> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a1sb = sext <2 x i8> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of 26 for: %a1sm = mul <2 x i32> %a1sa, %a1sb +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a1s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a1sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a2za = zext <4 x i8> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a2zb = zext <4 x i8> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2zm = mul <4 x i32> %a2za, %a2zb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a2sa = sext <4 x i8> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a2sb = sext <4 x i8> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2sm = mul <4 x i32> %a2sa, %a2sb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a2s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a2sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a3za = zext <8 x i8> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a3zb = zext <8 x i8> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a3zm = mul <8 x i32> %a3za, %a3zb +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a3z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a3sa = sext <8 x i8> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a3sb = sext <8 x i8> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a3sm = mul <8 x i32> %a3sa, %a3sb +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a3s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a3sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a4za = zext <16 x i8> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a4zb = zext <16 x i8> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a4zm = mul <16 x i32> %a4za, %a4zb +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a4z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a4sa = sext <16 x i8> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a4sb = sext <16 x i8> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a4sm = mul <16 x i32> %a4sa, %a4sb +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a4s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a4sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a5za = zext <1 x i16> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a5zb = zext <1 x i16> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a5zm = mul <1 x i32> %a5za, %a5zb +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a5z = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a5sa = sext <1 x i16> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a5sb = sext <1 x i16> undef to <1 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a5sm = mul <1 x i32> %a5sa, %a5sb +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a5s = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a5sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a6za = zext <2 x i16> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %a6zb = zext <2 x i16> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of 26 for: %a6zm = mul <2 x i32> %a6za, %a6zb +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a6z = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a6sa = sext <2 x i16> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %a6sb = sext <2 x i16> undef to <2 x i32> +; CHECK-NEXT: Cost Model: Found costs of 26 for: %a6sm = mul <2 x i32> %a6sa, %a6sb +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a6s = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a6sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7za = zext <4 x i16> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7zb = zext <4 x i16> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7zm = mul <4 x i32> %a7za, %a7zb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7z = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7sa = sext <4 x i16> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7sb = sext <4 x i16> undef to <4 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7sm = mul <4 x i32> %a7sa, %a7sb +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a7s = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a7sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a8za = zext <8 x i16> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a8zb = zext <8 x i16> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a8zm = mul <8 x i32> %a8za, %a8zb +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a8z = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a8sa = sext <8 x i16> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %a8sb = sext <8 x i16> undef to <8 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a8sm = mul <8 x i32> %a8sa, %a8sb +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a8s = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a8sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:40 CodeSize:1 Lat:1 SizeLat:1 for: %a9za = zext <16 x i16> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:40 CodeSize:1 Lat:1 SizeLat:1 for: %a9zb = zext <16 x i16> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a9zm = mul <16 x i32> %a9za, %a9zb +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a9z = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:40 CodeSize:1 Lat:1 SizeLat:1 for: %a9sa = sext <16 x i16> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:40 CodeSize:1 Lat:1 SizeLat:1 for: %a9sb = sext <16 x i16> undef to <16 x i32> +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a9sm = mul <16 x i32> %a9sa, %a9sb +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a9s = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a9sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a10m = mul <1 x i32> undef, undef +; CHECK-NEXT: Cost Model: Found costs of 4 for: %a10 = call i32 @llvm.vector.reduce.add.v1i32(<1 x i32> %a10m) +; CHECK-NEXT: Cost Model: Found costs of 10 for: %a11m = mul <2 x i32> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:54 CodeSize:38 Lat:54 SizeLat:54 for: %a11 = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %a11m) +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a12m = mul <4 x i32> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:2 SizeLat:2 for: %a12 = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %a12m) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a13m = mul <8 x i32> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a13 = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %a13m) +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a14m = mul <16 x i32> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %a14 = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %a14m) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0za = zext <1 x i8> undef to <1 x i32> %a0zb = zext <1 x i8> undef to <1 x i32> @@ -759,137 +759,137 @@ define void @mla_i32() { define void @mla_i64() { ; CHECK-LABEL: 'mla_i64' -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a0za = zext <1 x i8> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a0zb = zext <1 x i8> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0zm = mul <1 x i64> %a0za, %a0zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a0z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a0sa = sext <1 x i8> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a0sb = sext <1 x i8> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a0sm = mul <1 x i64> %a0sa, %a0sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a0s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1za = zext <2 x i8> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a1zb = zext <2 x i8> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %a1zm = mul <2 x i64> %a1za, %a1zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a1z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a1sa = sext <2 x i8> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a1sb = sext <2 x i8> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %a1sm = mul <2 x i64> %a1sa, %a1sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a1s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a2za = zext <4 x i8> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a2zb = zext <4 x i8> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 104 for instruction: %a2zm = mul <4 x i64> %a2za, %a2zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a2z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %a2sa = sext <4 x i8> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %a2sb = sext <4 x i8> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 104 for instruction: %a2sm = mul <4 x i64> %a2sa, %a2sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a2s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a3za = zext <8 x i8> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a3zb = zext <8 x i8> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a3zm = mul <8 x i64> %a3za, %a3zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a3z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 330 for instruction: %a3sa = sext <8 x i8> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 330 for instruction: %a3sb = sext <8 x i8> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a3sm = mul <8 x i64> %a3sa, %a3sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a3s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 298 for instruction: %a4za = zext <16 x i8> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 298 for instruction: %a4zb = zext <16 x i8> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 416 for instruction: %a4zm = mul <16 x i64> %a4za, %a4zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a4z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 1322 for instruction: %a4sa = sext <16 x i8> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 1322 for instruction: %a4sb = sext <16 x i8> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 416 for instruction: %a4sm = mul <16 x i64> %a4sa, %a4sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a4s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a5za = zext <1 x i16> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a5zb = zext <1 x i16> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5zm = mul <1 x i64> %a5za, %a5zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a5z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %a5sa = sext <1 x i16> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %a5sb = sext <1 x i16> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a5sm = mul <1 x i64> %a5sa, %a5sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a5s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a6za = zext <2 x i16> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a6zb = zext <2 x i16> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %a6zm = mul <2 x i64> %a6za, %a6zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a6z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a6sa = sext <2 x i16> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a6sb = sext <2 x i16> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %a6sm = mul <2 x i64> %a6sa, %a6sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a6s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a7za = zext <4 x i16> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a7zb = zext <4 x i16> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 104 for instruction: %a7zm = mul <4 x i64> %a7za, %a7zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a7z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %a7sa = sext <4 x i16> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %a7sb = sext <4 x i16> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 104 for instruction: %a7sm = mul <4 x i64> %a7sa, %a7sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a7s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a8za = zext <8 x i16> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %a8zb = zext <8 x i16> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a8zm = mul <8 x i64> %a8za, %a8zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a8z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 330 for instruction: %a8sa = sext <8 x i16> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 330 for instruction: %a8sb = sext <8 x i16> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a8sm = mul <8 x i64> %a8sa, %a8sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a8s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 296 for instruction: %a9za = zext <16 x i16> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 296 for instruction: %a9zb = zext <16 x i16> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 416 for instruction: %a9zm = mul <16 x i64> %a9za, %a9zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a9z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 1320 for instruction: %a9sa = sext <16 x i16> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 1320 for instruction: %a9sb = sext <16 x i16> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 416 for instruction: %a9sm = mul <16 x i64> %a9sa, %a9sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a9s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a10za = zext <1 x i32> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a10zb = zext <1 x i32> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a10zm = mul <1 x i64> %a10za, %a10zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a10z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a10sa = sext <1 x i32> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %a10sb = sext <1 x i32> undef to <1 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a10sm = mul <1 x i64> %a10sa, %a10sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a10s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a11za = zext <2 x i32> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a11zb = zext <2 x i32> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %a11zm = mul <2 x i64> %a11za, %a11zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a11z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %a11sa = sext <2 x i32> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %a11sb = sext <2 x i32> undef to <2 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %a11sm = mul <2 x i64> %a11sa, %a11sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a11s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a12za = zext <4 x i32> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %a12zb = zext <4 x i32> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 104 for instruction: %a12zm = mul <4 x i64> %a12za, %a12zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a12z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %a12sa = sext <4 x i32> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 66 for instruction: %a12sb = sext <4 x i32> undef to <4 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 104 for instruction: %a12sm = mul <4 x i64> %a12sa, %a12sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a12s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %a13za = zext <8 x i32> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %a13zb = zext <8 x i32> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a13zm = mul <8 x i64> %a13za, %a13zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a13z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 264 for instruction: %a13sa = sext <8 x i32> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 264 for instruction: %a13sb = sext <8 x i32> undef to <8 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a13sm = mul <8 x i64> %a13sa, %a13sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a13s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 288 for instruction: %a14za = zext <16 x i32> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 288 for instruction: %a14zb = zext <16 x i32> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 416 for instruction: %a14zm = mul <16 x i64> %a14za, %a14zb -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a14z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14zm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 1056 for instruction: %a14sa = sext <16 x i32> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 1056 for instruction: %a14sb = sext <16 x i32> undef to <16 x i64> -; CHECK-NEXT: Cost Model: Found an estimated cost of 416 for instruction: %a14sm = mul <16 x i64> %a14sa, %a14sb -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a14s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14sm) -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %a15m = mul <1 x i64> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %a15 = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a15m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %a16m = mul <2 x i64> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %a16 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a16m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %a17m = mul <4 x i64> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 208 for instruction: %a17 = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a17m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %a18m = mul <8 x i64> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 408 for instruction: %a18 = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a18m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %a19m = mul <16 x i64> undef, undef -; CHECK-NEXT: Cost Model: Found an estimated cost of 808 for instruction: %a19 = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a19m) -; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a0za = zext <1 x i8> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a0zb = zext <1 x i8> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a0zm = mul <1 x i64> %a0za, %a0zb +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a0z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a0sa = sext <1 x i8> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a0sb = sext <1 x i8> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a0sm = mul <1 x i64> %a0sa, %a0sb +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a0s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a0sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a1za = zext <2 x i8> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a1zb = zext <2 x i8> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of 52 for: %a1zm = mul <2 x i64> %a1za, %a1zb +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a1z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:10 Lat:20 SizeLat:20 for: %a1sa = sext <2 x i8> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:10 Lat:20 SizeLat:20 for: %a1sb = sext <2 x i8> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of 52 for: %a1sm = mul <2 x i64> %a1sa, %a1sb +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a1s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a1sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a2za = zext <4 x i8> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a2zb = zext <4 x i8> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of 104 for: %a2zm = mul <4 x i64> %a2za, %a2zb +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a2z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:82 CodeSize:1 Lat:1 SizeLat:1 for: %a2sa = sext <4 x i8> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:82 CodeSize:1 Lat:1 SizeLat:1 for: %a2sb = sext <4 x i8> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of 104 for: %a2sm = mul <4 x i64> %a2sa, %a2sb +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a2s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a2sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a3za = zext <8 x i8> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a3zb = zext <8 x i8> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of 208 for: %a3zm = mul <8 x i64> %a3za, %a3zb +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a3z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:330 CodeSize:1 Lat:1 SizeLat:1 for: %a3sa = sext <8 x i8> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:330 CodeSize:1 Lat:1 SizeLat:1 for: %a3sb = sext <8 x i8> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of 208 for: %a3sm = mul <8 x i64> %a3sa, %a3sb +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a3s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a3sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:298 CodeSize:1 Lat:1 SizeLat:1 for: %a4za = zext <16 x i8> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:298 CodeSize:1 Lat:1 SizeLat:1 for: %a4zb = zext <16 x i8> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of 416 for: %a4zm = mul <16 x i64> %a4za, %a4zb +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a4z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:1322 CodeSize:1 Lat:1 SizeLat:1 for: %a4sa = sext <16 x i8> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:1322 CodeSize:1 Lat:1 SizeLat:1 for: %a4sb = sext <16 x i8> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of 416 for: %a4sm = mul <16 x i64> %a4sa, %a4sb +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a4s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a4sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a5za = zext <1 x i16> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a5zb = zext <1 x i16> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a5zm = mul <1 x i64> %a5za, %a5zb +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a5z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:36 CodeSize:1 Lat:1 SizeLat:1 for: %a5sa = sext <1 x i16> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:36 CodeSize:1 Lat:1 SizeLat:1 for: %a5sb = sext <1 x i16> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a5sm = mul <1 x i64> %a5sa, %a5sb +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a5s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a5sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a6za = zext <2 x i16> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a6zb = zext <2 x i16> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of 52 for: %a6zm = mul <2 x i64> %a6za, %a6zb +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a6z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:10 Lat:20 SizeLat:20 for: %a6sa = sext <2 x i16> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:10 Lat:20 SizeLat:20 for: %a6sb = sext <2 x i16> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of 52 for: %a6sm = mul <2 x i64> %a6sa, %a6sb +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a6s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a6sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a7za = zext <4 x i16> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a7zb = zext <4 x i16> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of 104 for: %a7zm = mul <4 x i64> %a7za, %a7zb +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a7z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:82 CodeSize:1 Lat:1 SizeLat:1 for: %a7sa = sext <4 x i16> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:82 CodeSize:1 Lat:1 SizeLat:1 for: %a7sb = sext <4 x i16> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of 104 for: %a7sm = mul <4 x i64> %a7sa, %a7sb +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a7s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a7sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a8za = zext <8 x i16> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:74 CodeSize:1 Lat:1 SizeLat:1 for: %a8zb = zext <8 x i16> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of 208 for: %a8zm = mul <8 x i64> %a8za, %a8zb +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a8z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:330 CodeSize:1 Lat:1 SizeLat:1 for: %a8sa = sext <8 x i16> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:330 CodeSize:1 Lat:1 SizeLat:1 for: %a8sb = sext <8 x i16> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of 208 for: %a8sm = mul <8 x i64> %a8sa, %a8sb +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a8s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a8sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:296 CodeSize:1 Lat:1 SizeLat:1 for: %a9za = zext <16 x i16> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:296 CodeSize:1 Lat:1 SizeLat:1 for: %a9zb = zext <16 x i16> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of 416 for: %a9zm = mul <16 x i64> %a9za, %a9zb +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a9z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:1320 CodeSize:1 Lat:1 SizeLat:1 for: %a9sa = sext <16 x i16> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:1320 CodeSize:1 Lat:1 SizeLat:1 for: %a9sb = sext <16 x i16> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of 416 for: %a9sm = mul <16 x i64> %a9sa, %a9sb +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a9s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a9sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a10za = zext <1 x i32> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a10zb = zext <1 x i32> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a10zm = mul <1 x i64> %a10za, %a10zb +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a10z = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a10sa = sext <1 x i32> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:1 Lat:1 SizeLat:1 for: %a10sb = sext <1 x i32> undef to <1 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a10sm = mul <1 x i64> %a10sa, %a10sb +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a10s = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a10sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a11za = zext <2 x i32> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a11zb = zext <2 x i32> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of 52 for: %a11zm = mul <2 x i64> %a11za, %a11zb +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a11z = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %a11sa = sext <2 x i32> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %a11sb = sext <2 x i32> undef to <2 x i64> +; CHECK-NEXT: Cost Model: Found costs of 52 for: %a11sm = mul <2 x i64> %a11sa, %a11sb +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a11s = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a11sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a12za = zext <4 x i32> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %a12zb = zext <4 x i32> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of 104 for: %a12zm = mul <4 x i64> %a12za, %a12zb +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a12z = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:66 CodeSize:1 Lat:1 SizeLat:1 for: %a12sa = sext <4 x i32> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:66 CodeSize:1 Lat:1 SizeLat:1 for: %a12sb = sext <4 x i32> undef to <4 x i64> +; CHECK-NEXT: Cost Model: Found costs of 104 for: %a12sm = mul <4 x i64> %a12sa, %a12sb +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a12s = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a12sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:72 CodeSize:1 Lat:1 SizeLat:1 for: %a13za = zext <8 x i32> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:72 CodeSize:1 Lat:1 SizeLat:1 for: %a13zb = zext <8 x i32> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of 208 for: %a13zm = mul <8 x i64> %a13za, %a13zb +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a13z = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:264 CodeSize:1 Lat:1 SizeLat:1 for: %a13sa = sext <8 x i32> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:264 CodeSize:1 Lat:1 SizeLat:1 for: %a13sb = sext <8 x i32> undef to <8 x i64> +; CHECK-NEXT: Cost Model: Found costs of 208 for: %a13sm = mul <8 x i64> %a13sa, %a13sb +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a13s = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a13sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:288 CodeSize:1 Lat:1 SizeLat:1 for: %a14za = zext <16 x i32> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:288 CodeSize:1 Lat:1 SizeLat:1 for: %a14zb = zext <16 x i32> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of 416 for: %a14zm = mul <16 x i64> %a14za, %a14zb +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a14z = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14zm) +; CHECK-NEXT: Cost Model: Found costs of RThru:1056 CodeSize:1 Lat:1 SizeLat:1 for: %a14sa = sext <16 x i32> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of RThru:1056 CodeSize:1 Lat:1 SizeLat:1 for: %a14sb = sext <16 x i32> undef to <16 x i64> +; CHECK-NEXT: Cost Model: Found costs of 416 for: %a14sm = mul <16 x i64> %a14sa, %a14sb +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a14s = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a14sm) +; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %a15m = mul <1 x i64> undef, undef +; CHECK-NEXT: Cost Model: Found costs of 8 for: %a15 = call i64 @llvm.vector.reduce.add.v1i64(<1 x i64> %a15m) +; CHECK-NEXT: Cost Model: Found costs of 20 for: %a16m = mul <2 x i64> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:108 CodeSize:76 Lat:108 SizeLat:108 for: %a16 = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %a16m) +; CHECK-NEXT: Cost Model: Found costs of 40 for: %a17m = mul <4 x i64> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:208 CodeSize:144 Lat:208 SizeLat:208 for: %a17 = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %a17m) +; CHECK-NEXT: Cost Model: Found costs of 80 for: %a18m = mul <8 x i64> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:408 CodeSize:280 Lat:408 SizeLat:408 for: %a18 = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %a18m) +; CHECK-NEXT: Cost Model: Found costs of 160 for: %a19m = mul <16 x i64> undef, undef +; CHECK-NEXT: Cost Model: Found costs of RThru:808 CodeSize:552 Lat:808 SizeLat:808 for: %a19 = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %a19m) +; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void ; %a0za = zext <1 x i8> undef to <1 x i64> %a0zb = zext <1 x i8> undef to <1 x i64> |