Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[SveEmitter] Add builtins for svqadd, svqsub and svdot
This patch adds builtins for saturating add/sub instructions: - svqadd, svqadd_n - svqsub, svqsub_n and builtins for dot product instructions: - svdot, svdot_lane
- Loading branch information
1 parent
447efdb
commit 91cb13f
Showing
5 changed files
with
386 additions
and
5 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
147 changes: 147 additions & 0 deletions
147
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,147 @@ | ||
// RUN: %clang_cc1 -D__ARM_FEATURE_SVE -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s | ||
// RUN: %clang_cc1 -D__ARM_FEATURE_SVE -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s | ||
|
||
#include <arm_sve.h> | ||
|
||
#ifdef SVE_OVERLOADED_FORMS | ||
// A simple used,unused... macro, long enough to represent any SVE builtin. | ||
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3 | ||
#else | ||
#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4 | ||
#endif | ||
|
||
svint8_t test_svqadd_s8(svint8_t op1, svint8_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_s8 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> %op1, <vscale x 16 x i8> %op2) | ||
// CHECK: ret <vscale x 16 x i8> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_s8,,)(op1, op2); | ||
} | ||
|
||
svint16_t test_svqadd_s16(svint16_t op1, svint16_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_s16 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> %op1, <vscale x 8 x i16> %op2) | ||
// CHECK: ret <vscale x 8 x i16> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_s16,,)(op1, op2); | ||
} | ||
|
||
svint32_t test_svqadd_s32(svint32_t op1, svint32_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_s32 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> %op1, <vscale x 4 x i32> %op2) | ||
// CHECK: ret <vscale x 4 x i32> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_s32,,)(op1, op2); | ||
} | ||
|
||
svint64_t test_svqadd_s64(svint64_t op1, svint64_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_s64 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> %op1, <vscale x 2 x i64> %op2) | ||
// CHECK: ret <vscale x 2 x i64> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_s64,,)(op1, op2); | ||
} | ||
|
||
svuint8_t test_svqadd_u8(svuint8_t op1, svuint8_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_u8 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> %op1, <vscale x 16 x i8> %op2) | ||
// CHECK: ret <vscale x 16 x i8> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_u8,,)(op1, op2); | ||
} | ||
|
||
svuint16_t test_svqadd_u16(svuint16_t op1, svuint16_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_u16 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> %op1, <vscale x 8 x i16> %op2) | ||
// CHECK: ret <vscale x 8 x i16> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_u16,,)(op1, op2); | ||
} | ||
|
||
svuint32_t test_svqadd_u32(svuint32_t op1, svuint32_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_u32 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> %op1, <vscale x 4 x i32> %op2) | ||
// CHECK: ret <vscale x 4 x i32> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_u32,,)(op1, op2); | ||
} | ||
|
||
svuint64_t test_svqadd_u64(svuint64_t op1, svuint64_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_u64 | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> %op1, <vscale x 2 x i64> %op2) | ||
// CHECK: ret <vscale x 2 x i64> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_u64,,)(op1, op2); | ||
} | ||
|
||
svint8_t test_svqadd_n_s8(svint8_t op1, int8_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_s8 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> %op1, <vscale x 16 x i8> %[[DUP]]) | ||
// CHECK: ret <vscale x 16 x i8> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_s8,,)(op1, op2); | ||
} | ||
|
||
svint16_t test_svqadd_n_s16(svint16_t op1, int16_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_s16 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> %op1, <vscale x 8 x i16> %[[DUP]]) | ||
// CHECK: ret <vscale x 8 x i16> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_s16,,)(op1, op2); | ||
} | ||
|
||
svint32_t test_svqadd_n_s32(svint32_t op1, int32_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_s32 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> %op1, <vscale x 4 x i32> %[[DUP]]) | ||
// CHECK: ret <vscale x 4 x i32> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_s32,,)(op1, op2); | ||
} | ||
|
||
svint64_t test_svqadd_n_s64(svint64_t op1, int64_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_s64 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> %op1, <vscale x 2 x i64> %[[DUP]]) | ||
// CHECK: ret <vscale x 2 x i64> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_s64,,)(op1, op2); | ||
} | ||
|
||
svuint8_t test_svqadd_n_u8(svuint8_t op1, uint8_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_u8 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> %op1, <vscale x 16 x i8> %[[DUP]]) | ||
// CHECK: ret <vscale x 16 x i8> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_u8,,)(op1, op2); | ||
} | ||
|
||
svuint16_t test_svqadd_n_u16(svuint16_t op1, uint16_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_u16 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> %op1, <vscale x 8 x i16> %[[DUP]]) | ||
// CHECK: ret <vscale x 8 x i16> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_u16,,)(op1, op2); | ||
} | ||
|
||
svuint32_t test_svqadd_n_u32(svuint32_t op1, uint32_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_u32 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> %op1, <vscale x 4 x i32> %[[DUP]]) | ||
// CHECK: ret <vscale x 4 x i32> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_u32,,)(op1, op2); | ||
} | ||
|
||
svuint64_t test_svqadd_n_u64(svuint64_t op1, uint64_t op2) | ||
{ | ||
// CHECK-LABEL: test_svqadd_n_u64 | ||
// CHECK: %[[DUP:.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 %op2) | ||
// CHECK: %[[INTRINSIC:.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> %op1, <vscale x 2 x i64> %[[DUP]]) | ||
// CHECK: ret <vscale x 2 x i64> %[[INTRINSIC]] | ||
return SVE_ACLE_FUNC(svqadd,_n_u64,,)(op1, op2); | ||
} |
Oops, something went wrong.