Skip to content

[CIR][CIRGen][Builtin][Neon] Lower neon_vrnd32x #1388

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
Feb 24, 2025
Merged

[CIR][CIRGen][Builtin][Neon] Lower neon_vrnd32x #1388

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
14 changes: 8 additions & 6 deletions clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2619,6 +2619,14 @@ mlir::Value CIRGenFunction::emitCommonNeonBuiltinExpr(
: "aarch64.neon.srhadd";
break;
}
case NEON::BI__builtin_neon_vrnd32x_f32:
case NEON::BI__builtin_neon_vrnd32xq_f32:
case NEON::BI__builtin_neon_vrnd32x_f64:
case NEON::BI__builtin_neon_vrnd32xq_f64: {
intrincsName = "aarch64.neon.frint32x";
argTypes.push_back(vTy);
break;
}
case NEON::BI__builtin_neon_vshl_v:
case NEON::BI__builtin_neon_vshlq_v: {
return builder.create<cir::ShiftOp>(
Expand Down Expand Up @@ -4178,12 +4186,6 @@ CIRGenFunction::emitAArch64BuiltinExpr(unsigned BuiltinID, const CallExpr *E,
case NEON::BI__builtin_neon_vrndh_f16: {
llvm_unreachable("NEON::BI__builtin_neon_vrndh_f16 NYI");
}
case NEON::BI__builtin_neon_vrnd32x_f32:
case NEON::BI__builtin_neon_vrnd32xq_f32:
case NEON::BI__builtin_neon_vrnd32x_f64:
case NEON::BI__builtin_neon_vrnd32xq_f64: {
llvm_unreachable("NEON::BI__builtin_neon_vrnd32xq_f64 NYI");
}
case NEON::BI__builtin_neon_vrnd32z_f32:
case NEON::BI__builtin_neon_vrnd32zq_f32:
case NEON::BI__builtin_neon_vrnd32z_f64:
Expand Down
150 changes: 150 additions & 0 deletions clang/test/CIR/CodeGen/AArch64/v8.5a-neon-frint3264-intrinsic.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,150 @@
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +fullfp16 -target-feature +v8.5a \
// RUN: -fclangir -disable-O0-optnone \
// RUN: -flax-vector-conversions=none -emit-cir -o %t.cir %s
// RUN: FileCheck --check-prefix=CIR --input-file=%t.cir %s

// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +fullfp16 -target-feature +v8.5a \
// RUN: -fclangir -disable-O0-optnone \
// RUN: -flax-vector-conversions=none -emit-llvm -fno-clangir-call-conv-lowering -o - %s \
// RUN: | opt -S -passes=mem2reg,simplifycfg -o %t.ll
// RUN: FileCheck --check-prefix=LLVM --input-file=%t.ll %s

// REQUIRES: aarch64-registered-target || arm-registered-target

// This test mimics clang/test/CodeGen/AArch64/v8.2a-neon-frint3264-intrinsics.c, which eventually
// CIR shall be able to support fully. Since this is going to take some time to converge,
// the unsupported/NYI code is commented out, so that we can incrementally improve this.
// The NYI filecheck used contains the LLVM output from OG codegen that should guide the
// correct result when implementing this into the CIR pipeline.

#include <arm_neon.h>

float32x2_t test_vrnd32x_f32(float32x2_t a) {
return vrnd32x_f32(a);

// CIR-LABEL: vrnd32x_f32
// CIR: [[TMP0:%.*]] = cir.llvm.intrinsic "aarch64.neon.frint32x" {{.*}} : (!cir.vector<!cir.float x 2>) -> !cir.vector<!cir.float x 2>

// LLVM-LABEL: @test_vrnd32x_f32
// LLVM: [[RND:%.*]] = call <2 x float> @llvm.aarch64.neon.frint32x.v2f32(<2 x float> %0)
// LLVM: ret <2 x float> [[RND]]
}


float32x4_t test_vrnd32xq_f32(float32x4_t a) {
return vrnd32xq_f32(a);

// CIR-LABEL: vrnd32xq_f32
// CIR: [[TMP0:%.*]] = cir.llvm.intrinsic "aarch64.neon.frint32x" {{.*}} : (!cir.vector<!cir.float x 4>) -> !cir.vector<!cir.float x 4>

// LLVM-LABEL: @test_vrnd32xq_f32
// LLVM: [[RND:%.*]] = call <4 x float> @llvm.aarch64.neon.frint32x.v4f32(<4 x float> %0)
// LLVM: ret <4 x float> [[RND]]
}

// CHECK-LABEL: test_vrnd32z_f32
// CHECK: [[RND:%.*]] = call <2 x float> @llvm.aarch64.neon.frint32z.v2f32(<2 x float> %a)
// CHECK: ret <2 x float> [[RND]]
// float32x2_t test_vrnd32z_f32(float32x2_t a) {
// return vrnd32z_f32(a);
// }

// CHECK-LABEL: test_vrnd32zq_f32
// CHECK: [[RND:%.*]] = call <4 x float> @llvm.aarch64.neon.frint32z.v4f32(<4 x float> %a)
// CHECK: ret <4 x float> [[RND]]
// float32x4_t test_vrnd32zq_f32(float32x4_t a) {
// return vrnd32zq_f32(a);
// }

// CHECK-LABEL: test_vrnd64x_f32
// CHECK: [[RND:%.*]] = call <2 x float> @llvm.aarch64.neon.frint64x.v2f32(<2 x float> %a)
// CHECK: ret <2 x float> [[RND]]
// float32x2_t test_vrnd64x_f32(float32x2_t a) {
// return vrnd64x_f32(a);
// }

// CHECK-LABEL: test_vrnd64xq_f32
// CHECK: [[RND:%.*]] = call <4 x float> @llvm.aarch64.neon.frint64x.v4f32(<4 x float> %a)
// CHECK: ret <4 x float> [[RND]]
// float32x4_t test_vrnd64xq_f32(float32x4_t a) {
// return vrnd64xq_f32(a);
// }

// CHECK-LABEL: test_vrnd64z_f32
// CHECK: [[RND:%.*]] = call <2 x float> @llvm.aarch64.neon.frint64z.v2f32(<2 x float> %a)
// CHECK: ret <2 x float> [[RND]]
// float32x2_t test_vrnd64z_f32(float32x2_t a) {
// return vrnd64z_f32(a);
// }

// CHECK-LABEL: test_vrnd64zq_f32
// CHECK: [[RND:%.*]] = call <4 x float> @llvm.aarch64.neon.frint64z.v4f32(<4 x float> %a)
// CHECK: ret <4 x float> [[RND]]
// float32x4_t test_vrnd64zq_f32(float32x4_t a) {
// return vrnd64zq_f32(a);
// }

float64x1_t test_vrnd32x_f64(float64x1_t a) {
return vrnd32x_f64(a);

// CIR-LABEL: vrnd32x_f64
// CIR: [[TMP0:%.*]] = cir.llvm.intrinsic "aarch64.neon.frint32x" {{.*}} : (!cir.vector<!cir.double x 1>) -> !cir.vector<!cir.double x 1>

// LLVM-LABEL: @test_vrnd32x_f64
// LLVM: [[RND:%.*]] = call <1 x double> @llvm.aarch64.neon.frint32x.v1f64(<1 x double> %0)
// LLVM: ret <1 x double> [[RND]]
}


float64x2_t test_vrnd32xq_f64(float64x2_t a) {
return vrnd32xq_f64(a);

// CIR-LABEL: vrnd32xq_f64
// CIR: [[TMP0:%.*]] = cir.llvm.intrinsic "aarch64.neon.frint32x" {{.*}} : (!cir.vector<!cir.double x 2>) -> !cir.vector<!cir.double x 2>

// LLVM-LABEL: @test_vrnd32xq_f64
// LLVM: [[RND:%.*]] = call <2 x double> @llvm.aarch64.neon.frint32x.v2f64(<2 x double> %0)
// LLVM: ret <2 x double> [[RND]]
}

// CHECK-LABEL: test_vrnd32z_f64
// CHECK: [[RND:%.*]] = call <1 x double> @llvm.aarch64.neon.frint32z.v1f64(<1 x double> %a)
// CHECK: ret <1 x double> [[RND]]
// float64x1_t test_vrnd32z_f64(float64x1_t a) {
// return vrnd32z_f64(a);
// }

// CHECK-LABEL: test_vrnd32zq_f64
// CHECK: [[RND:%.*]] = call <2 x double> @llvm.aarch64.neon.frint32z.v2f64(<2 x double> %a)
// CHECK: ret <2 x double> [[RND]]
// float64x2_t test_vrnd32zq_f64(float64x2_t a) {
// return vrnd32zq_f64(a);
// }

// CHECK-LABEL: test_vrnd64x_f64
// CHECK: [[RND:%.*]] = call <1 x double> @llvm.aarch64.neon.frint64x.v1f64(<1 x double> %a)
// CHECK: ret <1 x double> [[RND]]
// float64x1_t test_vrnd64x_f64(float64x1_t a) {
// return vrnd64x_f64(a);
// }

// CHECK-LABEL: test_vrnd64xq_f64
// CHECK: [[RND:%.*]] = call <2 x double> @llvm.aarch64.neon.frint64x.v2f64(<2 x double> %a)
// CHECK: ret <2 x double> [[RND]]
// float64x2_t test_vrnd64xq_f64(float64x2_t a) {
// return vrnd64xq_f64(a);
// }

// CHECK-LABEL: test_vrnd64z_f64
// CHECK: [[RND:%.*]] = call <1 x double> @llvm.aarch64.neon.frint64z.v1f64(<1 x double> %a)
// CHECK: ret <1 x double> [[RND]]
// float64x1_t test_vrnd64z_f64(float64x1_t a) {
// return vrnd64z_f64(a);
// }

// CHECK-LABEL: test_vrnd64zq_f64
// CHECK: [[RND:%.*]] = call <2 x double> @llvm.aarch64.neon.frint64z.v2f64(<2 x double> %a)
// CHECK: ret <2 x double> [[RND]]
// float64x2_t test_vrnd64zq_f64(float64x2_t a) {
// return vrnd64zq_f64(a);
// }
Loading