[PATCH] D134978: [RISCV] Update cost of vector roundeven to match round which uses the same sequence but a different FRM value.
Craig Topper via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Fri Sep 30 11:23:23 PDT 2022
craig.topper created this revision.
craig.topper added reviewers: reames, eopXD.
Herald added subscribers: sunshaoce, VincentWu, StephenFan, vkmr, frasercrmck, evandro, luismarques, apazos, sameer.abuasal, s.egerton, Jim, benna, psnobl, jocewei, PkmX, the_o, brucehoult, MartinMosbeck, rogfer01, edward-jones, zzheng, jrtc27, shiva0217, kito-cheng, niosHD, sabuasal, simoncook, johnrusso, rbar, asb, hiraditya, arichardson.
Herald added a project: All.
craig.topper requested review of this revision.
Herald added subscribers: pcwang-thead, MaskRay.
Herald added a project: LLVM.
Repository:
rG LLVM Github Monorepo
https://reviews.llvm.org/D134978
Files:
llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
llvm/test/Analysis/CostModel/RISCV/fround.ll
Index: llvm/test/Analysis/CostModel/RISCV/fround.ll
===================================================================
--- llvm/test/Analysis/CostModel/RISCV/fround.ll
+++ llvm/test/Analysis/CostModel/RISCV/fround.ll
@@ -262,23 +262,23 @@
define void @roundeven() {
; CHECK-LABEL: 'roundeven'
; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %1 = call float @llvm.roundeven.f32(float undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %2 = call <2 x float> @llvm.roundeven.v2f32(<2 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %3 = call <4 x float> @llvm.roundeven.v4f32(<4 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %4 = call <8 x float> @llvm.roundeven.v8f32(<8 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %5 = call <16 x float> @llvm.roundeven.v16f32(<16 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %6 = call <vscale x 2 x float> @llvm.roundeven.nxv2f32(<vscale x 2 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %7 = call <vscale x 4 x float> @llvm.roundeven.nxv4f32(<vscale x 4 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %8 = call <vscale x 8 x float> @llvm.roundeven.nxv8f32(<vscale x 8 x float> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %9 = call <vscale x 16 x float> @llvm.roundeven.nxv16f32(<vscale x 16 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %2 = call <2 x float> @llvm.roundeven.v2f32(<2 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %3 = call <4 x float> @llvm.roundeven.v4f32(<4 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %4 = call <8 x float> @llvm.roundeven.v8f32(<8 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %5 = call <16 x float> @llvm.roundeven.v16f32(<16 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %6 = call <vscale x 2 x float> @llvm.roundeven.nxv2f32(<vscale x 2 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %7 = call <vscale x 4 x float> @llvm.roundeven.nxv4f32(<vscale x 4 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %8 = call <vscale x 8 x float> @llvm.roundeven.nxv8f32(<vscale x 8 x float> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %9 = call <vscale x 16 x float> @llvm.roundeven.nxv16f32(<vscale x 16 x float> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %10 = call double @llvm.roundeven.f64(double undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %11 = call <2 x double> @llvm.roundeven.v2f64(<2 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %12 = call <4 x double> @llvm.roundeven.v4f64(<4 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %13 = call <8 x double> @llvm.roundeven.v8f64(<8 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %14 = call <16 x double> @llvm.roundeven.v16f64(<16 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %15 = call <vscale x 1 x double> @llvm.roundeven.nxv1f64(<vscale x 1 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %16 = call <vscale x 2 x double> @llvm.roundeven.nxv2f64(<vscale x 2 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %17 = call <vscale x 4 x double> @llvm.roundeven.nxv4f64(<vscale x 4 x double> undef)
-; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %18 = call <vscale x 8 x double> @llvm.roundeven.nxv8f64(<vscale x 8 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %11 = call <2 x double> @llvm.roundeven.v2f64(<2 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %12 = call <4 x double> @llvm.roundeven.v4f64(<4 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %13 = call <8 x double> @llvm.roundeven.v8f64(<8 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %14 = call <16 x double> @llvm.roundeven.v16f64(<16 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %15 = call <vscale x 1 x double> @llvm.roundeven.nxv1f64(<vscale x 1 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %16 = call <vscale x 2 x double> @llvm.roundeven.nxv2f64(<vscale x 2 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %17 = call <vscale x 4 x double> @llvm.roundeven.nxv4f64(<vscale x 4 x double> undef)
+; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %18 = call <vscale x 8 x double> @llvm.roundeven.nxv8f64(<vscale x 8 x double> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
;
call float @llvm.roundeven.f32(float undef)
Index: llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
===================================================================
--- llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -328,6 +328,22 @@
{Intrinsic::round, MVT::nxv2f64, 9},
{Intrinsic::round, MVT::nxv4f64, 9},
{Intrinsic::round, MVT::nxv8f64, 9},
+ {Intrinsic::roundeven, MVT::v2f32, 9},
+ {Intrinsic::roundeven, MVT::v4f32, 9},
+ {Intrinsic::roundeven, MVT::v8f32, 9},
+ {Intrinsic::roundeven, MVT::v16f32, 9},
+ {Intrinsic::roundeven, MVT::nxv2f32, 9},
+ {Intrinsic::roundeven, MVT::nxv4f32, 9},
+ {Intrinsic::roundeven, MVT::nxv8f32, 9},
+ {Intrinsic::roundeven, MVT::nxv16f32, 9},
+ {Intrinsic::roundeven, MVT::v2f64, 9},
+ {Intrinsic::roundeven, MVT::v4f64, 9},
+ {Intrinsic::roundeven, MVT::v8f64, 9},
+ {Intrinsic::roundeven, MVT::v16f64, 9},
+ {Intrinsic::roundeven, MVT::nxv1f64, 9},
+ {Intrinsic::roundeven, MVT::nxv2f64, 9},
+ {Intrinsic::roundeven, MVT::nxv4f64, 9},
+ {Intrinsic::roundeven, MVT::nxv8f64, 9},
{Intrinsic::fabs, MVT::v2f32, 1},
{Intrinsic::fabs, MVT::v4f32, 1},
{Intrinsic::fabs, MVT::v8f32, 1},
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D134978.464336.patch
Type: text/x-patch
Size: 6673 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220930/649874bf/attachment.bin>
More information about the llvm-commits
mailing list