Skip to content

Commit

Permalink
[CIR][CIRGen][Builtin][Neon] Lower neon_vshlq_v (#1042)
Browse files Browse the repository at this point in the history
  • Loading branch information
@ghehg @lanza
ghehg authored and lanza committed Nov 4, 2024
1 parent 9228ff9 commit adf0c99
Show file tree
Hide file tree
Showing 2 changed files with 111 additions and 66 deletions.
5 changes: 5 additions & 0 deletions clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2452,6 +2452,11 @@ mlir::Value CIRGenFunction::buildCommonNeonBuiltinExpr(
: "llvm.aarch64.neon.srhadd";
break;
}
case NEON::BI__builtin_neon_vshlq_v: {
intrincsName = (intrinicId != altLLVMIntrinsic) ? "llvm.aarch64.neon.ushl"
: "llvm.aarch64.neon.sshl";
break;
}
case NEON::BI__builtin_neon_vhadd_v:
case NEON::BI__builtin_neon_vhaddq_v: {
intrincsName = (intrinicId != altLLVMIntrinsic) ? "llvm.aarch64.neon.uhadd"
Expand Down
172 changes: 106 additions & 66 deletions clang/test/CIR/CodeGen/AArch64/neon.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3405,79 +3405,119 @@ int8x8_t test_vqadd_s8(int8x8_t a, int8x8_t b) {
// return vshl_u64(a, b);
// }

// NYI-LABEL: @test_vshlq_s8(
// NYI: [[VSHLQ_V_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.sshl.v16i8(<16 x i8> %a, <16 x i8> %b)
// NYI: ret <16 x i8> [[VSHLQ_V_I]]
// int8x16_t test_vshlq_s8(int8x16_t a, int8x16_t b) {
// return vshlq_s8(a, b);
// }
int8x16_t test_vshlq_s8(int8x16_t a, int8x16_t b) {
return vshlq_s8(a, b);

// NYI-LABEL: @test_vshlq_s16(
// NYI: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8>
// NYI: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8>
// NYI: [[VSHLQ_V2_I:%.*]] = call <8 x i16> @llvm.aarch64.neon.sshl.v8i16(<8 x i16> %a, <8 x i16> %b)
// NYI: [[VSHLQ_V3_I:%.*]] = bitcast <8 x i16> [[VSHLQ_V2_I]] to <16 x i8>
// NYI: ret <8 x i16> [[VSHLQ_V2_I]]
// int16x8_t test_vshlq_s16(int16x8_t a, int16x8_t b) {
// return vshlq_s16(a, b);
// }
// CIR-LABEL: vshlq_s8
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.sshl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!s8i x 16>, !cir.vector<!s8i x 16>) -> !cir.vector<!s8i x 16>

// NYI-LABEL: @test_vshlq_s32(
// NYI: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8>
// NYI: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8>
// NYI: [[VSHLQ_V2_I:%.*]] = call <4 x i32> @llvm.aarch64.neon.sshl.v4i32(<4 x i32> %a, <4 x i32> %b)
// NYI: [[VSHLQ_V3_I:%.*]] = bitcast <4 x i32> [[VSHLQ_V2_I]] to <16 x i8>
// NYI: ret <4 x i32> [[VSHLQ_V2_I]]
// int32x4_t test_vshlq_s32(int32x4_t a, int32x4_t b) {
// return vshlq_s32(a, b);
// }
// LLVM: {{.*}}test_vshlq_s8(<16 x i8>{{.*}}[[A:%.*]], <16 x i8>{{.*}}[[B:%.*]])
// LLVM: [[VSHLQ_V_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.sshl.v16i8(<16 x i8> [[A]], <16 x i8> [[B]])
// LLVM: ret <16 x i8> [[VSHLQ_V_I]]
}

// NYI-LABEL: @test_vshlq_s64(
// NYI: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
// NYI: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8>
// NYI: [[VSHLQ_V2_I:%.*]] = call <2 x i64> @llvm.aarch64.neon.sshl.v2i64(<2 x i64> %a, <2 x i64> %b)
// NYI: [[VSHLQ_V3_I:%.*]] = bitcast <2 x i64> [[VSHLQ_V2_I]] to <16 x i8>
// NYI: ret <2 x i64> [[VSHLQ_V2_I]]
// int64x2_t test_vshlq_s64(int64x2_t a, int64x2_t b) {
// return vshlq_s64(a, b);
// }
int16x8_t test_vshlq_s16(int16x8_t a, int16x8_t b) {
return vshlq_s16(a, b);

// NYI-LABEL: @test_vshlq_u8(
// NYI: [[VSHLQ_V_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.ushl.v16i8(<16 x i8> %a, <16 x i8> %b)
// NYI: ret <16 x i8> [[VSHLQ_V_I]]
// uint8x16_t test_vshlq_u8(uint8x16_t a, int8x16_t b) {
// return vshlq_u8(a, b);
// }
// CIR-LABEL: vshlq_s16
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.sshl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!s16i x 8>, !cir.vector<!s16i x 8>) -> !cir.vector<!s16i x 8>

// NYI-LABEL: @test_vshlq_u16(
// NYI: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8>
// NYI: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8>
// NYI: [[VSHLQ_V2_I:%.*]] = call <8 x i16> @llvm.aarch64.neon.ushl.v8i16(<8 x i16> %a, <8 x i16> %b)
// NYI: [[VSHLQ_V3_I:%.*]] = bitcast <8 x i16> [[VSHLQ_V2_I]] to <16 x i8>
// NYI: ret <8 x i16> [[VSHLQ_V2_I]]
// uint16x8_t test_vshlq_u16(uint16x8_t a, int16x8_t b) {
// return vshlq_u16(a, b);
// }
// LLVM: {{.*}}test_vshlq_s16(<8 x i16>{{.*}}[[A:%.*]], <8 x i16>{{.*}}[[B:%.*]])
// LLVM: [[TMP0:%.*]] = bitcast <8 x i16> [[A]] to <16 x i8>
// LLVM: [[TMP1:%.*]] = bitcast <8 x i16> [[B]] to <16 x i8>
// LLVM: [[VSHLQ_V2_I:%.*]] = call <8 x i16> @llvm.aarch64.neon.sshl.v8i16(<8 x i16> [[A]], <8 x i16> [[B]])
// LLVM: [[VSHLQ_V3_I:%.*]] = bitcast <8 x i16> [[VSHLQ_V2_I]] to <16 x i8>
// LLVM: ret <8 x i16> [[VSHLQ_V2_I]]
}

// NYI-LABEL: @test_vshlq_u32(
// NYI: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8>
// NYI: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8>
// NYI: [[VSHLQ_V2_I:%.*]] = call <4 x i32> @llvm.aarch64.neon.ushl.v4i32(<4 x i32> %a, <4 x i32> %b)
// NYI: [[VSHLQ_V3_I:%.*]] = bitcast <4 x i32> [[VSHLQ_V2_I]] to <16 x i8>
// NYI: ret <4 x i32> [[VSHLQ_V2_I]]
// uint32x4_t test_vshlq_u32(uint32x4_t a, int32x4_t b) {
// return vshlq_u32(a, b);
// }
int32x4_t test_vshlq_s32(int32x4_t a, int32x4_t b) {
return vshlq_s32(a, b);

// NYI-LABEL: @test_vshlq_u64(
// NYI: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
// NYI: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8>
// NYI: [[VSHLQ_V2_I:%.*]] = call <2 x i64> @llvm.aarch64.neon.ushl.v2i64(<2 x i64> %a, <2 x i64> %b)
// NYI: [[VSHLQ_V3_I:%.*]] = bitcast <2 x i64> [[VSHLQ_V2_I]] to <16 x i8>
// NYI: ret <2 x i64> [[VSHLQ_V2_I]]
// uint64x2_t test_vshlq_u64(uint64x2_t a, int64x2_t b) {
// return vshlq_u64(a, b);
// }
// CIR-LABEL: vshlq_s32
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.sshl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!s32i x 4>, !cir.vector<!s32i x 4>) -> !cir.vector<!s32i x 4>

// LLVM: {{.*}}test_vshlq_s32(<4 x i32>{{.*}}[[A:%.*]], <4 x i32>{{.*}}[[B:%.*]])
// LLVM: [[TMP0:%.*]] = bitcast <4 x i32> [[A]] to <16 x i8>
// LLVM: [[TMP1:%.*]] = bitcast <4 x i32> [[B]] to <16 x i8>
// LLVM: [[VSHLQ_V2_I:%.*]] = call <4 x i32> @llvm.aarch64.neon.sshl.v4i32(<4 x i32> [[A]], <4 x i32> [[B]])
// LLVM: [[VSHLQ_V3_I:%.*]] = bitcast <4 x i32> [[VSHLQ_V2_I]] to <16 x i8>
// LLVM: ret <4 x i32> [[VSHLQ_V2_I]]
}

int64x2_t test_vshlq_s64(int64x2_t a, int64x2_t b) {
return vshlq_s64(a, b);

// CIR-LABEL: vshlq_s64
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.sshl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!s64i x 2>, !cir.vector<!s64i x 2>) -> !cir.vector<!s64i x 2>

// LLVM: {{.*}}test_vshlq_s64(<2 x i64>{{.*}}[[A:%.*]], <2 x i64>{{.*}}[[B:%.*]])
// LLVM: [[TMP0:%.*]] = bitcast <2 x i64> [[A]] to <16 x i8>
// LLVM: [[TMP1:%.*]] = bitcast <2 x i64> [[B]] to <16 x i8>
// LLVM: [[VSHLQ_V2_I:%.*]] = call <2 x i64> @llvm.aarch64.neon.sshl.v2i64(<2 x i64> [[A]], <2 x i64> [[B]])
// LLVM: [[VSHLQ_V3_I:%.*]] = bitcast <2 x i64> [[VSHLQ_V2_I]] to <16 x i8>
// LLVM: ret <2 x i64> [[VSHLQ_V2_I]]
}

uint8x16_t test_vshlq_u8(uint8x16_t a, int8x16_t b) {
return vshlq_u8(a, b);

// CIR-LABEL: vshlq_u8
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.ushl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!u8i x 16>, !cir.vector<!u8i x 16>) -> !cir.vector<!u8i x 16>

// LLVM: {{.*}}test_vshlq_u8(<16 x i8>{{.*}}[[A:%.*]], <16 x i8>{{.*}}[[B:%.*]])
// LLVM: [[VSHLQ_V_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.ushl.v16i8(<16 x i8> [[A]], <16 x i8> [[B]])
// LLVM: ret <16 x i8> [[VSHLQ_V_I]]
}

uint16x8_t test_vshlq_u16(uint16x8_t a, int16x8_t b) {
return vshlq_u16(a, b);

// CIR-LABEL: vshlq_u16
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.ushl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!u16i x 8>, !cir.vector<!u16i x 8>) -> !cir.vector<!u16i x 8>

// LLVM: {{.*}}test_vshlq_u16(<8 x i16>{{.*}}[[A:%.*]], <8 x i16>{{.*}}[[B:%.*]])
// LLVM: [[TMP0:%.*]] = bitcast <8 x i16> [[A]] to <16 x i8>
// LLVM: [[TMP1:%.*]] = bitcast <8 x i16> [[B]] to <16 x i8>
// LLVM: [[VSHLQ_V2_I:%.*]] = call <8 x i16> @llvm.aarch64.neon.ushl.v8i16(<8 x i16> [[A]], <8 x i16> [[B]])
// LLVM: [[VSHLQ_V3_I:%.*]] = bitcast <8 x i16> [[VSHLQ_V2_I]] to <16 x i8>
// LLVM: ret <8 x i16> [[VSHLQ_V2_I]]
}

uint32x4_t test_vshlq_u32(uint32x4_t a, int32x4_t b) {
return vshlq_u32(a, b);

// CIR-LABEL: vshlq_u32
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.ushl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!u32i x 4>, !cir.vector<!u32i x 4>) -> !cir.vector<!u32i x 4>

// LLVM: {{.*}}test_vshlq_u32(<4 x i32>{{.*}}[[A:%.*]], <4 x i32>{{.*}}[[B:%.*]])
// LLVM: [[TMP0:%.*]] = bitcast <4 x i32> [[A]] to <16 x i8>
// LLVM: [[TMP1:%.*]] = bitcast <4 x i32> [[B]] to <16 x i8>
// LLVM: [[VSHLQ_V2_I:%.*]] = call <4 x i32> @llvm.aarch64.neon.ushl.v4i32(<4 x i32> [[A]], <4 x i32> [[B]])
// LLVM: [[VSHLQ_V3_I:%.*]] = bitcast <4 x i32> [[VSHLQ_V2_I]] to <16 x i8>
// LLVM: ret <4 x i32> [[VSHLQ_V2_I]]
}

uint64x2_t test_vshlq_u64(uint64x2_t a, int64x2_t b) {
return vshlq_u64(a, b);

// CIR-LABEL: vshlq_u64
// CIR: {{%.*}} = cir.llvm.intrinsic "llvm.aarch64.neon.ushl" {{%.*}}, {{%.*}} :
// CIR-SAME: (!cir.vector<!u64i x 2>, !cir.vector<!u64i x 2>) -> !cir.vector<!u64i x 2>

// LLVM: {{.*}}test_vshlq_u64(<2 x i64>{{.*}}[[A:%.*]], <2 x i64>{{.*}}[[B:%.*]])
// LLVM: [[TMP0:%.*]] = bitcast <2 x i64> [[A]] to <16 x i8>
// LLVM: [[TMP1:%.*]] = bitcast <2 x i64> [[B]] to <16 x i8>
// LLVM: [[VSHLQ_V2_I:%.*]] = call <2 x i64> @llvm.aarch64.neon.ushl.v2i64(<2 x i64> [[A]], <2 x i64> [[B]])
// LLVM: [[VSHLQ_V3_I:%.*]] = bitcast <2 x i64> [[VSHLQ_V2_I]] to <16 x i8>
// LLVM: ret <2 x i64> [[VSHLQ_V2_I]]
}

// NYI-LABEL: @test_vqshl_s8(
// NYI: [[VQSHL_V_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshl.v8i8(<8 x i8> %a, <8 x i8> %b)
Expand Down

0 comments on commit adf0c99

Please sign in to comment.