Skip to content

[AArch64] Update costs for scalarizing i64->f32 int_to_fp. #132366

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Mar 26, 2025
Merged

[AArch64] Update costs for scalarizing i64->f32 int_to_fp. #132366

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 14 additions & 2 deletions llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2939,10 +2939,8 @@ InstructionCost AArch64TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
// Complex: to v2f32
{ISD::SINT_TO_FP, MVT::v2f32, MVT::v2i8, 3},
{ISD::SINT_TO_FP, MVT::v2f32, MVT::v2i16, 3},
{ISD::SINT_TO_FP, MVT::v2f32, MVT::v2i64, 2},
{ISD::UINT_TO_FP, MVT::v2f32, MVT::v2i8, 3},
{ISD::UINT_TO_FP, MVT::v2f32, MVT::v2i16, 3},
{ISD::UINT_TO_FP, MVT::v2f32, MVT::v2i64, 2},

// Complex: to v4f32
{ISD::SINT_TO_FP, MVT::v4f32, MVT::v4i8, 4},
Expand Down Expand Up @@ -3199,6 +3197,20 @@ InstructionCost AArch64TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
FP16Tbl, ISD, DstTy.getSimpleVT(), SrcTy.getSimpleVT()))
return AdjustCost(Entry->Cost);

// INT_TO_FP of i64->f32 will scalarize, which is required to avoid
// double-rounding issues.
if ((ISD == ISD::SINT_TO_FP || ISD == ISD::UINT_TO_FP) &&
DstTy.getScalarType() == MVT::f32 && SrcTy.getScalarSizeInBits() > 32 &&
Copy link
Contributor

@pranavk pranavk Mar 24, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you want == 64 here instead of > 32?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Anything larger than a i32 (like an i48, if they ever come up) will be legalized to a i64 as we convert to legal types. I figure that anything larger than i64 will become an i128 which will be scalarized too, although the i64 is the important one.

isa<FixedVectorType>(Dst) && isa<FixedVectorType>(Src))
return AdjustCost(
cast<FixedVectorType>(Dst)->getNumElements() *
getCastInstrCost(Opcode, Dst->getScalarType(), Src->getScalarType(),
CCH, CostKind) +
BaseT::getScalarizationOverhead(cast<FixedVectorType>(Src), false, true,
CostKind) +
BaseT::getScalarizationOverhead(cast<FixedVectorType>(Dst), true, false,
CostKind));

if ((ISD == ISD::ZERO_EXTEND || ISD == ISD::SIGN_EXTEND) &&
CCH == TTI::CastContextHint::Masked &&
ST->isSVEorStreamingSVEAvailable() &&
Expand Down
16 changes: 8 additions & 8 deletions llvm/test/Analysis/CostModel/AArch64/cast.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -391,8 +391,8 @@ define i32 @casts_no_users() {
; CHECK-NEXT: Cost Model: Found costs of RThru:3 CodeSize:1 Lat:1 SizeLat:1 for: %r175 = sitofp <2 x i16> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of 1 for: %r176 = uitofp <2 x i32> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of 1 for: %r177 = sitofp <2 x i32> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %r178 = uitofp <2 x i64> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %r179 = sitofp <2 x i64> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:1 Lat:1 SizeLat:1 for: %r178 = uitofp <2 x i64> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:1 Lat:1 SizeLat:1 for: %r179 = sitofp <2 x i64> undef to <2 x float>
; CHECK-NEXT: Cost Model: Found costs of 1 for: %r180 = uitofp <2 x i1> undef to <2 x double>
; CHECK-NEXT: Cost Model: Found costs of 1 for: %r181 = sitofp <2 x i1> undef to <2 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %r182 = uitofp <2 x i8> undef to <2 x double>
Expand All @@ -411,8 +411,8 @@ define i32 @casts_no_users() {
; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %r195 = sitofp <4 x i16> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of 1 for: %r196 = uitofp <4 x i32> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of 1 for: %r197 = sitofp <4 x i32> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:5 CodeSize:1 Lat:1 SizeLat:1 for: %r198 = uitofp <4 x i64> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:5 CodeSize:1 Lat:1 SizeLat:1 for: %r199 = sitofp <4 x i64> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %r198 = uitofp <4 x i64> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:18 CodeSize:1 Lat:1 SizeLat:1 for: %r199 = sitofp <4 x i64> undef to <4 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:3 CodeSize:1 Lat:1 SizeLat:1 for: %r200 = uitofp <4 x i1> undef to <4 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:3 CodeSize:1 Lat:1 SizeLat:1 for: %r201 = sitofp <4 x i1> undef to <4 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:9 CodeSize:1 Lat:1 SizeLat:1 for: %r202 = uitofp <4 x i8> undef to <4 x double>
Expand All @@ -431,8 +431,8 @@ define i32 @casts_no_users() {
; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %r215 = sitofp <8 x i16> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %r216 = uitofp <8 x i32> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:2 CodeSize:1 Lat:1 SizeLat:1 for: %r217 = sitofp <8 x i32> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %r218 = uitofp <8 x i64> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:10 CodeSize:1 Lat:1 SizeLat:1 for: %r219 = sitofp <8 x i64> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:36 CodeSize:1 Lat:1 SizeLat:1 for: %r218 = uitofp <8 x i64> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:36 CodeSize:1 Lat:1 SizeLat:1 for: %r219 = sitofp <8 x i64> undef to <8 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:1 Lat:1 SizeLat:1 for: %r220 = uitofp <8 x i1> undef to <8 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:1 Lat:1 SizeLat:1 for: %r221 = sitofp <8 x i1> undef to <8 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:19 CodeSize:1 Lat:1 SizeLat:1 for: %r222 = uitofp <8 x i8> undef to <8 x double>
Expand All @@ -451,8 +451,8 @@ define i32 @casts_no_users() {
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:1 Lat:1 SizeLat:1 for: %r235 = sitofp <16 x i16> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %r236 = uitofp <16 x i32> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:1 Lat:1 SizeLat:1 for: %r237 = sitofp <16 x i32> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:1 Lat:1 SizeLat:1 for: %r238 = uitofp <16 x i64> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:20 CodeSize:1 Lat:1 SizeLat:1 for: %r239 = sitofp <16 x i64> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:72 CodeSize:1 Lat:1 SizeLat:1 for: %r238 = uitofp <16 x i64> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:72 CodeSize:1 Lat:1 SizeLat:1 for: %r239 = sitofp <16 x i64> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found costs of RThru:15 CodeSize:1 Lat:1 SizeLat:1 for: %r240 = uitofp <16 x i1> undef to <16 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:15 CodeSize:1 Lat:1 SizeLat:1 for: %r241 = sitofp <16 x i1> undef to <16 x double>
; CHECK-NEXT: Cost Model: Found costs of RThru:39 CodeSize:1 Lat:1 SizeLat:1 for: %r242 = uitofp <16 x i8> undef to <16 x double>
Expand Down
14 changes: 9 additions & 5 deletions llvm/test/Analysis/CostModel/AArch64/no-sve-no-neon.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,18 @@
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 2
; RUN: opt -mattr=-neon < %s -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s
; RUN: opt -mattr=+sve,-neon < %s -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s
; RUN: opt -mattr=-neon < %s -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=CHECK-NONEON
; RUN: opt -mattr=+sve,-neon < %s -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefixes=CHECK-WITHSVE

target triple = "aarch64-unknown-linux-gnu"
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"

define void @uitofp() {
; CHECK-LABEL: 'uitofp'
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %conv = uitofp <16 x i64> undef to <16 x float>
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
; CHECK-NONEON-LABEL: 'uitofp'
; CHECK-NONEON-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %conv = uitofp <16 x i64> undef to <16 x float>
; CHECK-NONEON-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-WITHSVE-LABEL: 'uitofp'
; CHECK-WITHSVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %conv = uitofp <16 x i64> undef to <16 x float>
; CHECK-WITHSVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
%conv = uitofp <16 x i64> undef to <16 x float>
ret void
Expand Down
24 changes: 12 additions & 12 deletions llvm/test/Analysis/CostModel/AArch64/sve-cast.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -696,8 +696,8 @@ define i32 @casts_no_users() {
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %r175 = sitofp <2 x i16> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r176 = uitofp <2 x i32> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r177 = sitofp <2 x i32> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r178 = uitofp <2 x i64> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r179 = sitofp <2 x i64> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r178 = uitofp <2 x i64> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r179 = sitofp <2 x i64> undef to <2 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r180 = uitofp <2 x i1> undef to <2 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r181 = sitofp <2 x i1> undef to <2 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %r182 = uitofp <2 x i8> undef to <2 x double>
Expand All @@ -716,8 +716,8 @@ define i32 @casts_no_users() {
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r195 = sitofp <4 x i16> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r196 = uitofp <4 x i32> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r197 = sitofp <4 x i32> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %r198 = uitofp <4 x i64> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %r199 = sitofp <4 x i64> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %r198 = uitofp <4 x i64> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %r199 = sitofp <4 x i64> undef to <4 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %r200 = uitofp <4 x i1> undef to <4 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %r201 = sitofp <4 x i1> undef to <4 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %r202 = uitofp <4 x i8> undef to <4 x double>
Expand All @@ -736,8 +736,8 @@ define i32 @casts_no_users() {
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %r215 = sitofp <8 x i16> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r216 = uitofp <8 x i32> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r217 = sitofp <8 x i32> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %r218 = uitofp <8 x i64> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %r219 = sitofp <8 x i64> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %r218 = uitofp <8 x i64> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %r219 = sitofp <8 x i64> undef to <8 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %r220 = uitofp <8 x i1> undef to <8 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %r221 = sitofp <8 x i1> undef to <8 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %r222 = uitofp <8 x i8> undef to <8 x double>
Expand All @@ -756,8 +756,8 @@ define i32 @casts_no_users() {
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r235 = sitofp <16 x i16> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %r236 = uitofp <16 x i32> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %r237 = sitofp <16 x i32> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %r238 = uitofp <16 x i64> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %r239 = sitofp <16 x i64> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %r238 = uitofp <16 x i64> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %r239 = sitofp <16 x i64> undef to <16 x float>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %r240 = uitofp <16 x i1> undef to <16 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %r241 = sitofp <16 x i1> undef to <16 x double>
; CHECK-SVE-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %r242 = uitofp <16 x i8> undef to <16 x double>
Expand Down Expand Up @@ -1122,8 +1122,8 @@ define i32 @casts_no_users() {
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %r175 = sitofp <2 x i16> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r176 = uitofp <2 x i32> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r177 = sitofp <2 x i32> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r178 = uitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r179 = sitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r178 = uitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r179 = sitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r180 = uitofp <2 x i1> undef to <2 x double>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r181 = sitofp <2 x i1> undef to <2 x double>
; FIXED-MIN-256-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %r182 = uitofp <2 x i8> undef to <2 x double>
Expand Down Expand Up @@ -1335,8 +1335,8 @@ define i32 @casts_no_users() {
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %r175 = sitofp <2 x i16> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r176 = uitofp <2 x i32> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r177 = sitofp <2 x i32> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r178 = uitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %r179 = sitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r178 = uitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %r179 = sitofp <2 x i64> undef to <2 x float>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r180 = uitofp <2 x i1> undef to <2 x double>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %r181 = sitofp <2 x i1> undef to <2 x double>
; FIXED-MIN-2048-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %r182 = uitofp <2 x i8> undef to <2 x double>
Expand Down
Loading