Skip to content

[VectorUtils] Trivially vectorize ldexp, [l]lround #145545

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

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
+510 −0
Open

[VectorUtils] Trivially vectorize ldexp, [l]lround #145545

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6ce3e2a
[VectorUtils] Trivially vectorize ldexp, [l]lround
artagnon Jun 24, 2025
bff4172
[CostModel] Add tests for ldexp, [l]lround
artagnon Jun 25, 2025
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
6 changes: 6 additions & 0 deletions llvm/lib/Analysis/VectorUtils.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,6 +81,7 @@ bool llvm::isTriviallyVectorizable(Intrinsic::ID ID) {
case Intrinsic::exp:
case Intrinsic::exp10:
case Intrinsic::exp2:
case Intrinsic::ldexp:
case Intrinsic::log:
case Intrinsic::log10:
case Intrinsic::log2:
Expand Down Expand Up @@ -108,6 +109,8 @@ bool llvm::isTriviallyVectorizable(Intrinsic::ID ID) {
case Intrinsic::canonicalize:
case Intrinsic::fptosi_sat:
case Intrinsic::fptoui_sat:
case Intrinsic::lround:
case Intrinsic::llround:
case Intrinsic::lrint:
case Intrinsic::llrint:
case Intrinsic::ucmp:
Expand Down Expand Up @@ -189,6 +192,8 @@ bool llvm::isVectorIntrinsicWithOverloadTypeAtArg(
switch (ID) {
case Intrinsic::fptosi_sat:
case Intrinsic::fptoui_sat:
case Intrinsic::lround:
case Intrinsic::llround:
case Intrinsic::lrint:
case Intrinsic::llrint:
case Intrinsic::vp_lrint:
Expand All @@ -203,6 +208,7 @@ bool llvm::isVectorIntrinsicWithOverloadTypeAtArg(
case Intrinsic::vp_is_fpclass:
return OpdIdx == 0;
case Intrinsic::powi:
case Intrinsic::ldexp:
return OpdIdx == -1 || OpdIdx == 1;
default:
return OpdIdx == -1;
Expand Down
109 changes: 109 additions & 0 deletions llvm/test/Analysis/CostModel/RISCV/exp.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,109 @@
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
Copy link
Contributor

Choose a reason for hiding this comment

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

Thanks for adding the extra tests, I think you can commit the tests separately.

; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -S -mtriple=riscv64 -mattr=+v,+f,+d,+zvfh,+zvfbfmin %s | FileCheck %s

define void @ldexp() {
; CHECK-LABEL: 'ldexp'
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %1 = call bfloat @llvm.ldexp.bf16.i32(bfloat poison, i32 poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %2 = call <2 x bfloat> @llvm.ldexp.v2bf16.v2i32(<2 x bfloat> poison, <2 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %3 = call <4 x bfloat> @llvm.ldexp.v4bf16.v4i32(<4 x bfloat> poison, <4 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %4 = call <8 x bfloat> @llvm.ldexp.v8bf16.v8i32(<8 x bfloat> poison, <8 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %5 = call <16 x bfloat> @llvm.ldexp.v16bf16.v16i32(<16 x bfloat> poison, <16 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %6 = call <vscale x 1 x bfloat> @llvm.ldexp.nxv1bf16.nxv1i32(<vscale x 1 x bfloat> poison, <vscale x 1 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %7 = call <vscale x 2 x bfloat> @llvm.ldexp.nxv2bf16.nxv2i32(<vscale x 2 x bfloat> poison, <vscale x 2 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %8 = call <vscale x 4 x bfloat> @llvm.ldexp.nxv4bf16.nxv4i32(<vscale x 4 x bfloat> poison, <vscale x 4 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %9 = call <vscale x 8 x bfloat> @llvm.ldexp.nxv8bf16.nxv8i32(<vscale x 8 x bfloat> poison, <vscale x 8 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %10 = call <vscale x 16 x bfloat> @llvm.ldexp.nxv16bf16.nxv16i32(<vscale x 16 x bfloat> poison, <vscale x 16 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %11 = call float @llvm.ldexp.f32.i32(float poison, i32 poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %12 = call <2 x float> @llvm.ldexp.v2f32.v2i32(<2 x float> poison, <2 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %13 = call <4 x float> @llvm.ldexp.v4f32.v4i32(<4 x float> poison, <4 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %14 = call <8 x float> @llvm.ldexp.v8f32.v8i32(<8 x float> poison, <8 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %15 = call <16 x float> @llvm.ldexp.v16f32.v16i32(<16 x float> poison, <16 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %16 = call <vscale x 1 x float> @llvm.ldexp.nxv1f32.nxv1i32(<vscale x 1 x float> poison, <vscale x 1 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %17 = call <vscale x 2 x float> @llvm.ldexp.nxv2f32.nxv2i32(<vscale x 2 x float> poison, <vscale x 2 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %18 = call <vscale x 4 x float> @llvm.ldexp.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %19 = call <vscale x 8 x float> @llvm.ldexp.nxv8f32.nxv8i32(<vscale x 8 x float> poison, <vscale x 8 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %20 = call <vscale x 16 x float> @llvm.ldexp.nxv16f32.nxv16i32(<vscale x 16 x float> poison, <vscale x 16 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %21 = call double @llvm.ldexp.f64.i32(double poison, i32 poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %22 = call <2 x double> @llvm.ldexp.v2f64.v2i32(<2 x double> poison, <2 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %23 = call <4 x double> @llvm.ldexp.v4f64.v4i32(<4 x double> poison, <4 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %24 = call <8 x double> @llvm.ldexp.v8f64.v8i32(<8 x double> poison, <8 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %25 = call <16 x double> @llvm.ldexp.v16f64.v16i32(<16 x double> poison, <16 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %26 = call <vscale x 1 x double> @llvm.ldexp.nxv1f64.nxv1i32(<vscale x 1 x double> poison, <vscale x 1 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %27 = call <vscale x 2 x double> @llvm.ldexp.nxv2f64.nxv2i32(<vscale x 2 x double> poison, <vscale x 2 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %28 = call <vscale x 4 x double> @llvm.ldexp.nxv4f64.nxv4i32(<vscale x 4 x double> poison, <vscale x 4 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %29 = call <vscale x 8 x double> @llvm.ldexp.nxv8f64.nxv8i32(<vscale x 8 x double> poison, <vscale x 8 x i32> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %30 = call <vscale x 16 x double> @llvm.ldexp.nxv16f64.nxv16i32(<vscale x 16 x double> poison, <vscale x 16 x i32> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %31 = call float @llvm.ldexp.f32.i64(float poison, i64 poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %32 = call <2 x float> @llvm.ldexp.v2f32.v2i64(<2 x float> poison, <2 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %33 = call <4 x float> @llvm.ldexp.v4f32.v4i64(<4 x float> poison, <4 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %34 = call <8 x float> @llvm.ldexp.v8f32.v8i64(<8 x float> poison, <8 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %35 = call <16 x float> @llvm.ldexp.v16f32.v16i64(<16 x float> poison, <16 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %36 = call <vscale x 1 x float> @llvm.ldexp.nxv1f32.nxv1i64(<vscale x 1 x float> poison, <vscale x 1 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %37 = call <vscale x 2 x float> @llvm.ldexp.nxv2f32.nxv2i64(<vscale x 2 x float> poison, <vscale x 2 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %38 = call <vscale x 4 x float> @llvm.ldexp.nxv4f32.nxv4i64(<vscale x 4 x float> poison, <vscale x 4 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %39 = call <vscale x 8 x float> @llvm.ldexp.nxv8f32.nxv8i64(<vscale x 8 x float> poison, <vscale x 8 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %40 = call <vscale x 16 x float> @llvm.ldexp.nxv16f32.nxv16i64(<vscale x 16 x float> poison, <vscale x 16 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %41 = call double @llvm.ldexp.f64.i64(double poison, i64 poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %42 = call <2 x double> @llvm.ldexp.v2f64.v2i64(<2 x double> poison, <2 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %43 = call <4 x double> @llvm.ldexp.v4f64.v4i64(<4 x double> poison, <4 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %44 = call <8 x double> @llvm.ldexp.v8f64.v8i64(<8 x double> poison, <8 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %45 = call <16 x double> @llvm.ldexp.v16f64.v16i64(<16 x double> poison, <16 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %46 = call <vscale x 1 x double> @llvm.ldexp.nxv1f64.nxv1i64(<vscale x 1 x double> poison, <vscale x 1 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %47 = call <vscale x 2 x double> @llvm.ldexp.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %48 = call <vscale x 4 x double> @llvm.ldexp.nxv4f64.nxv4i64(<vscale x 4 x double> poison, <vscale x 4 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %49 = call <vscale x 8 x double> @llvm.ldexp.nxv8f64.nxv8i64(<vscale x 8 x double> poison, <vscale x 8 x i64> poison)
; CHECK-NEXT: Cost Model: Invalid cost for instruction: %50 = call <vscale x 16 x double> @llvm.ldexp.nxv16f64.nxv16i64(<vscale x 16 x double> poison, <vscale x 16 x i64> poison)
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
call bfloat @llvm.ldexp.bf16.i32(bfloat poison, i32 poison)
call <2 x bfloat> @llvm.ldexp.v2bf16.v2i32(<2 x bfloat> poison, <2 x i32> poison)
call <4 x bfloat> @llvm.ldexp.v4bf16.v4i32(<4 x bfloat> poison, <4 x i32> poison)
call <8 x bfloat> @llvm.ldexp.v8bf16.v8i32(<8 x bfloat> poison, <8 x i32> poison)
call <16 x bfloat> @llvm.ldexp.v16bf16.v16i32(<16 x bfloat> poison, <16 x i32> poison)
call <vscale x 1 x bfloat> @llvm.ldexp.nxv1bf16.nxv1i32(<vscale x 1 x bfloat> poison, <vscale x 1 x i32> poison)
call <vscale x 2 x bfloat> @llvm.ldexp.nxv2bf16.nxv2i32(<vscale x 2 x bfloat> poison, <vscale x 2 x i32> poison)
call <vscale x 4 x bfloat> @llvm.ldexp.nxv4bf16.nxv4i32(<vscale x 4 x bfloat> poison, <vscale x 4 x i32> poison)
call <vscale x 8 x bfloat> @llvm.ldexp.nxv8bf16.nxv8i32(<vscale x 8 x bfloat> poison, <vscale x 8 x i32> poison)
call <vscale x 16 x bfloat> @llvm.ldexp.nxv16bf16.nxv16i32(<vscale x 16 x bfloat> poison, <vscale x 16 x i32> poison)
call float @llvm.ldexp.f32.i32(float poison, i32 poison)
call <2 x float> @llvm.ldexp.v2f32.v2i32(<2 x float> poison, <2 x i32> poison)
call <4 x float> @llvm.ldexp.v4f32.v4i32(<4 x float> poison, <4 x i32> poison)
call <8 x float> @llvm.ldexp.v8f32.v8i32(<8 x float> poison, <8 x i32> poison)
call <16 x float> @llvm.ldexp.v16f32.v16i32(<16 x float> poison, <16 x i32> poison)
call <vscale x 1 x float> @llvm.ldexp.nxv1f32.nxv1i32(<vscale x 1 x float> poison, <vscale x 1 x i32> poison)
call <vscale x 2 x float> @llvm.ldexp.nxv2f32.nxv2i32(<vscale x 2 x float> poison, <vscale x 2 x i32> poison)
call <vscale x 4 x float> @llvm.ldexp.nxv4f32.nxv4i32(<vscale x 4 x float> poison, <vscale x 4 x i32> poison)
call <vscale x 8 x float> @llvm.ldexp.nxv8f32.nxv8i32(<vscale x 8 x float> poison, <vscale x 8 x i32> poison)
call <vscale x 16 x float> @llvm.ldexp.nxv16f32.nxv16i32(<vscale x 16 x float> poison, <vscale x 16 x i32> poison)
call double @llvm.ldexp.f64.i32(double poison, i32 poison)
call <2 x double> @llvm.ldexp.v2f64.v2i32(<2 x double> poison, <2 x i32> poison)
call <4 x double> @llvm.ldexp.v4f64.v4i32(<4 x double> poison, <4 x i32> poison)
call <8 x double> @llvm.ldexp.v8f64.v8i32(<8 x double> poison, <8 x i32> poison)
call <16 x double> @llvm.ldexp.v16f64.v16i32(<16 x double> poison, <16 x i32> poison)
call <vscale x 1 x double> @llvm.ldexp.nxv1f64.nxv1i32(<vscale x 1 x double> poison, <vscale x 1 x i32> poison)
call <vscale x 2 x double> @llvm.ldexp.nxv2f64.nxv2i32(<vscale x 2 x double> poison, <vscale x 2 x i32> poison)
call <vscale x 4 x double> @llvm.ldexp.nxv4f64.nxv4i32(<vscale x 4 x double> poison, <vscale x 4 x i32> poison)
call <vscale x 8 x double> @llvm.ldexp.nxv8f64.nxv8i32(<vscale x 8 x double> poison, <vscale x 8 x i32> poison)
call <vscale x 16 x double> @llvm.ldexp.nxv16f64.nxv16i32(<vscale x 16 x double> poison, <vscale x 16 x i32> poison)
call float @llvm.ldexp.f32.i64(float poison, i64 poison)
call <2 x float> @llvm.ldexp.v2f32.v2i64(<2 x float> poison, <2 x i64> poison)
call <4 x float> @llvm.ldexp.v4f32.v4i64(<4 x float> poison, <4 x i64> poison)
call <8 x float> @llvm.ldexp.v8f32.v8i64(<8 x float> poison, <8 x i64> poison)
call <16 x float> @llvm.ldexp.v16f32.v16i64(<16 x float> poison, <16 x i64> poison)
call <vscale x 1 x float> @llvm.ldexp.nxv1f32.nxv1i64(<vscale x 1 x float> poison, <vscale x 1 x i64> poison)
call <vscale x 2 x float> @llvm.ldexp.nxv2f32.nxv2i64(<vscale x 2 x float> poison, <vscale x 2 x i64> poison)
call <vscale x 4 x float> @llvm.ldexp.nxv4f32.nxv4i64(<vscale x 4 x float> poison, <vscale x 4 x i64> poison)
call <vscale x 8 x float> @llvm.ldexp.nxv8f32.nxv8i64(<vscale x 8 x float> poison, <vscale x 8 x i64> poison)
call <vscale x 16 x float> @llvm.ldexp.nxv16f32.nxv16i64(<vscale x 16 x float> poison, <vscale x 16 x i64> poison)
call double @llvm.ldexp.f64.i64(double poison, i64 poison)
call <2 x double> @llvm.ldexp.v2f64.v2i64(<2 x double> poison, <2 x i64> poison)
call <4 x double> @llvm.ldexp.v4f64.v4i64(<4 x double> poison, <4 x i64> poison)
call <8 x double> @llvm.ldexp.v8f64.v8i64(<8 x double> poison, <8 x i64> poison)
call <16 x double> @llvm.ldexp.v16f64.v16i64(<16 x double> poison, <16 x i64> poison)
call <vscale x 1 x double> @llvm.ldexp.nxv1f64.nxv1i64(<vscale x 1 x double> poison, <vscale x 1 x i64> poison)
call <vscale x 2 x double> @llvm.ldexp.nxv2f64.nxv2i64(<vscale x 2 x double> poison, <vscale x 2 x i64> poison)
call <vscale x 4 x double> @llvm.ldexp.nxv4f64.nxv4i64(<vscale x 4 x double> poison, <vscale x 4 x i64> poison)
call <vscale x 8 x double> @llvm.ldexp.nxv8f64.nxv8i64(<vscale x 8 x double> poison, <vscale x 8 x i64> poison)
call <vscale x 16 x double> @llvm.ldexp.nxv16f64.nxv16i64(<vscale x 16 x double> poison, <vscale x 16 x i64> poison)
ret void
}
Loading