Changeset View
Changeset View
Standalone View
Standalone View
clang/test/CodeGen/aarch64-neon-intrinsics.c
- This file is larger than 256 KB, so syntax highlighting is disabled by default.
| // RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \ | // RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \ | ||||
| // RUN: -fallow-half-arguments-and-returns -S -disable-O0-optnone -emit-llvm -o - %s \ | // RUN: -fallow-half-arguments-and-returns -S -disable-O0-optnone \ | ||||
| // RUN: -flax-vector-conversions=none -emit-llvm -o - %s \ | |||||
| // RUN: | opt -S -mem2reg \ | // RUN: | opt -S -mem2reg \ | ||||
| // RUN: | FileCheck %s | // RUN: | FileCheck %s | ||||
| // Test new aarch64 intrinsics and types | // Test new aarch64 intrinsics and types | ||||
| #include <arm_neon.h> | #include <arm_neon.h> | ||||
| // CHECK-LABEL: @test_vadd_s8( | // CHECK-LABEL: @test_vadd_s8( | ||||
| ▲ Show 20 Lines • Show All 390 Lines • ▼ Show 20 Lines | |||||
| } | } | ||||
| // CHECK-LABEL: @test_vmla_s16( | // CHECK-LABEL: @test_vmla_s16( | ||||
| // CHECK: [[MUL_I:%.*]] = mul <4 x i16> %v2, %v3 | // CHECK: [[MUL_I:%.*]] = mul <4 x i16> %v2, %v3 | ||||
| // CHECK: [[ADD_I:%.*]] = add <4 x i16> %v1, [[MUL_I]] | // CHECK: [[ADD_I:%.*]] = add <4 x i16> %v1, [[MUL_I]] | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> [[ADD_I]] to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> [[ADD_I]] to <8 x i8> | ||||
| // CHECK: ret <8 x i8> [[TMP0]] | // CHECK: ret <8 x i8> [[TMP0]] | ||||
| int8x8_t test_vmla_s16(int16x4_t v1, int16x4_t v2, int16x4_t v3) { | int8x8_t test_vmla_s16(int16x4_t v1, int16x4_t v2, int16x4_t v3) { | ||||
| return vmla_s16(v1, v2, v3); | return (int8x8_t)vmla_s16(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vmla_s32( | // CHECK-LABEL: @test_vmla_s32( | ||||
| // CHECK: [[MUL_I:%.*]] = mul <2 x i32> %v2, %v3 | // CHECK: [[MUL_I:%.*]] = mul <2 x i32> %v2, %v3 | ||||
| // CHECK: [[ADD_I:%.*]] = add <2 x i32> %v1, [[MUL_I]] | // CHECK: [[ADD_I:%.*]] = add <2 x i32> %v1, [[MUL_I]] | ||||
| // CHECK: ret <2 x i32> [[ADD_I]] | // CHECK: ret <2 x i32> [[ADD_I]] | ||||
| int32x2_t test_vmla_s32(int32x2_t v1, int32x2_t v2, int32x2_t v3) { | int32x2_t test_vmla_s32(int32x2_t v1, int32x2_t v2, int32x2_t v3) { | ||||
| return vmla_s32(v1, v2, v3); | return vmla_s32(v1, v2, v3); | ||||
| ▲ Show 20 Lines • Show All 104 Lines • ▼ Show 20 Lines | |||||
| } | } | ||||
| // CHECK-LABEL: @test_vmls_s16( | // CHECK-LABEL: @test_vmls_s16( | ||||
| // CHECK: [[MUL_I:%.*]] = mul <4 x i16> %v2, %v3 | // CHECK: [[MUL_I:%.*]] = mul <4 x i16> %v2, %v3 | ||||
| // CHECK: [[SUB_I:%.*]] = sub <4 x i16> %v1, [[MUL_I]] | // CHECK: [[SUB_I:%.*]] = sub <4 x i16> %v1, [[MUL_I]] | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> [[SUB_I]] to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> [[SUB_I]] to <8 x i8> | ||||
| // CHECK: ret <8 x i8> [[TMP0]] | // CHECK: ret <8 x i8> [[TMP0]] | ||||
| int8x8_t test_vmls_s16(int16x4_t v1, int16x4_t v2, int16x4_t v3) { | int8x8_t test_vmls_s16(int16x4_t v1, int16x4_t v2, int16x4_t v3) { | ||||
| return vmls_s16(v1, v2, v3); | return (int8x8_t)vmls_s16(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vmls_s32( | // CHECK-LABEL: @test_vmls_s32( | ||||
| // CHECK: [[MUL_I:%.*]] = mul <2 x i32> %v2, %v3 | // CHECK: [[MUL_I:%.*]] = mul <2 x i32> %v2, %v3 | ||||
| // CHECK: [[SUB_I:%.*]] = sub <2 x i32> %v1, [[MUL_I]] | // CHECK: [[SUB_I:%.*]] = sub <2 x i32> %v1, [[MUL_I]] | ||||
| // CHECK: ret <2 x i32> [[SUB_I]] | // CHECK: ret <2 x i32> [[SUB_I]] | ||||
| int32x2_t test_vmls_s32(int32x2_t v1, int32x2_t v2, int32x2_t v3) { | int32x2_t test_vmls_s32(int32x2_t v1, int32x2_t v2, int32x2_t v3) { | ||||
| return vmls_s32(v1, v2, v3); | return vmls_s32(v1, v2, v3); | ||||
| ▲ Show 20 Lines • Show All 434 Lines • ▼ Show 20 Lines | |||||
| // CHECK: [[TMP2:%.*]] = bitcast <4 x i16> %v3 to <8 x i8> | // CHECK: [[TMP2:%.*]] = bitcast <4 x i16> %v3 to <8 x i8> | ||||
| // CHECK: [[VBSL3_I:%.*]] = and <4 x i16> %v1, %v2 | // CHECK: [[VBSL3_I:%.*]] = and <4 x i16> %v1, %v2 | ||||
| // CHECK: [[TMP3:%.*]] = xor <4 x i16> %v1, <i16 -1, i16 -1, i16 -1, i16 -1> | // CHECK: [[TMP3:%.*]] = xor <4 x i16> %v1, <i16 -1, i16 -1, i16 -1, i16 -1> | ||||
| // CHECK: [[VBSL4_I:%.*]] = and <4 x i16> [[TMP3]], %v3 | // CHECK: [[VBSL4_I:%.*]] = and <4 x i16> [[TMP3]], %v3 | ||||
| // CHECK: [[VBSL5_I:%.*]] = or <4 x i16> [[VBSL3_I]], [[VBSL4_I]] | // CHECK: [[VBSL5_I:%.*]] = or <4 x i16> [[VBSL3_I]], [[VBSL4_I]] | ||||
| // CHECK: [[TMP4:%.*]] = bitcast <4 x i16> [[VBSL5_I]] to <8 x i8> | // CHECK: [[TMP4:%.*]] = bitcast <4 x i16> [[VBSL5_I]] to <8 x i8> | ||||
| // CHECK: ret <8 x i8> [[TMP4]] | // CHECK: ret <8 x i8> [[TMP4]] | ||||
| int8x8_t test_vbsl_s16(uint16x4_t v1, int16x4_t v2, int16x4_t v3) { | int8x8_t test_vbsl_s16(uint16x4_t v1, int16x4_t v2, int16x4_t v3) { | ||||
| return vbsl_s16(v1, v2, v3); | return (int8x8_t)vbsl_s16(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vbsl_s32( | // CHECK-LABEL: @test_vbsl_s32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %v1 to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %v1 to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %v2 to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %v2 to <8 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <2 x i32> %v3 to <8 x i8> | // CHECK: [[TMP2:%.*]] = bitcast <2 x i32> %v3 to <8 x i8> | ||||
| // CHECK: [[VBSL3_I:%.*]] = and <2 x i32> %v1, %v2 | // CHECK: [[VBSL3_I:%.*]] = and <2 x i32> %v1, %v2 | ||||
| // CHECK: [[TMP3:%.*]] = xor <2 x i32> %v1, <i32 -1, i32 -1> | // CHECK: [[TMP3:%.*]] = xor <2 x i32> %v1, <i32 -1, i32 -1> | ||||
| // CHECK: [[VBSL4_I:%.*]] = and <2 x i32> [[TMP3]], %v3 | // CHECK: [[VBSL4_I:%.*]] = and <2 x i32> [[TMP3]], %v3 | ||||
| // CHECK: [[VBSL5_I:%.*]] = or <2 x i32> [[VBSL3_I]], [[VBSL4_I]] | // CHECK: [[VBSL5_I:%.*]] = or <2 x i32> [[VBSL3_I]], [[VBSL4_I]] | ||||
| // CHECK: ret <2 x i32> [[VBSL5_I]] | // CHECK: ret <2 x i32> [[VBSL5_I]] | ||||
| int32x2_t test_vbsl_s32(uint32x2_t v1, int32x2_t v2, int32x2_t v3) { | int32x2_t test_vbsl_s32(uint32x2_t v1, int32x2_t v2, int32x2_t v3) { | ||||
| return vbsl_s32(v1, v2, v3); | return vbsl_s32(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vbsl_s64( | // CHECK-LABEL: @test_vbsl_s64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %v1 to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %v1 to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %v2 to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %v2 to <8 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <1 x i64> %v3 to <8 x i8> | // CHECK: [[TMP2:%.*]] = bitcast <1 x i64> %v3 to <8 x i8> | ||||
| // CHECK: [[VBSL3_I:%.*]] = and <1 x i64> %v1, %v2 | // CHECK: [[VBSL3_I:%.*]] = and <1 x i64> %v1, %v2 | ||||
| // CHECK: [[TMP3:%.*]] = xor <1 x i64> %v1, <i64 -1> | // CHECK: [[TMP3:%.*]] = xor <1 x i64> %v1, <i64 -1> | ||||
| // CHECK: [[VBSL4_I:%.*]] = and <1 x i64> [[TMP3]], %v3 | // CHECK: [[VBSL4_I:%.*]] = and <1 x i64> [[TMP3]], %v3 | ||||
| // CHECK: [[VBSL5_I:%.*]] = or <1 x i64> [[VBSL3_I]], [[VBSL4_I]] | // CHECK: [[VBSL5_I:%.*]] = or <1 x i64> [[VBSL3_I]], [[VBSL4_I]] | ||||
| // CHECK: ret <1 x i64> [[VBSL5_I]] | // CHECK: ret <1 x i64> [[VBSL5_I]] | ||||
| uint64x1_t test_vbsl_s64(uint64x1_t v1, uint64x1_t v2, uint64x1_t v3) { | int64x1_t test_vbsl_s64(uint64x1_t v1, int64x1_t v2, int64x1_t v3) { | ||||
| return vbsl_s64(v1, v2, v3); | return vbsl_s64(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vbsl_u8( | // CHECK-LABEL: @test_vbsl_u8( | ||||
| // CHECK: [[VBSL_I:%.*]] = and <8 x i8> %v1, %v2 | // CHECK: [[VBSL_I:%.*]] = and <8 x i8> %v1, %v2 | ||||
| // CHECK: [[TMP0:%.*]] = xor <8 x i8> %v1, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> | // CHECK: [[TMP0:%.*]] = xor <8 x i8> %v1, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> | ||||
| // CHECK: [[VBSL1_I:%.*]] = and <8 x i8> [[TMP0]], %v3 | // CHECK: [[VBSL1_I:%.*]] = and <8 x i8> [[TMP0]], %v3 | ||||
| // CHECK: [[VBSL2_I:%.*]] = or <8 x i8> [[VBSL_I]], [[VBSL1_I]] | // CHECK: [[VBSL2_I:%.*]] = or <8 x i8> [[VBSL_I]], [[VBSL1_I]] | ||||
| Show All 37 Lines | |||||
| // CHECK: [[VBSL4_I:%.*]] = and <1 x i64> [[TMP3]], %v3 | // CHECK: [[VBSL4_I:%.*]] = and <1 x i64> [[TMP3]], %v3 | ||||
| // CHECK: [[VBSL5_I:%.*]] = or <1 x i64> [[VBSL3_I]], [[VBSL4_I]] | // CHECK: [[VBSL5_I:%.*]] = or <1 x i64> [[VBSL3_I]], [[VBSL4_I]] | ||||
| // CHECK: ret <1 x i64> [[VBSL5_I]] | // CHECK: ret <1 x i64> [[VBSL5_I]] | ||||
| uint64x1_t test_vbsl_u64(uint64x1_t v1, uint64x1_t v2, uint64x1_t v3) { | uint64x1_t test_vbsl_u64(uint64x1_t v1, uint64x1_t v2, uint64x1_t v3) { | ||||
| return vbsl_u64(v1, v2, v3); | return vbsl_u64(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vbsl_f32( | // CHECK-LABEL: @test_vbsl_f32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x float> %v1 to <2 x i32> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %v1 to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> [[TMP0]] to <8 x i8> | |||||
| // CHECK: [[TMP2:%.*]] = bitcast <2 x float> %v2 to <8 x i8> | // CHECK: [[TMP2:%.*]] = bitcast <2 x float> %v2 to <8 x i8> | ||||
| // CHECK: [[TMP3:%.*]] = bitcast <2 x float> %v3 to <8 x i8> | // CHECK: [[TMP3:%.*]] = bitcast <2 x float> %v3 to <8 x i8> | ||||
| // CHECK: [[VBSL1_I:%.*]] = bitcast <8 x i8> [[TMP2]] to <2 x i32> | // CHECK: [[VBSL1_I:%.*]] = bitcast <8 x i8> [[TMP2]] to <2 x i32> | ||||
| // CHECK: [[VBSL2_I:%.*]] = bitcast <8 x i8> [[TMP3]] to <2 x i32> | // CHECK: [[VBSL2_I:%.*]] = bitcast <8 x i8> [[TMP3]] to <2 x i32> | ||||
| // CHECK: [[VBSL3_I:%.*]] = and <2 x i32> [[TMP0]], [[VBSL1_I]] | // CHECK: [[VBSL3_I:%.*]] = and <2 x i32> %v1, [[VBSL1_I]] | ||||
| // CHECK: [[TMP4:%.*]] = xor <2 x i32> [[TMP0]], <i32 -1, i32 -1> | // CHECK: [[TMP4:%.*]] = xor <2 x i32> %v1, <i32 -1, i32 -1> | ||||
| // CHECK: [[VBSL4_I:%.*]] = and <2 x i32> [[TMP4]], [[VBSL2_I]] | // CHECK: [[VBSL4_I:%.*]] = and <2 x i32> [[TMP4]], [[VBSL2_I]] | ||||
| // CHECK: [[VBSL5_I:%.*]] = or <2 x i32> [[VBSL3_I]], [[VBSL4_I]] | // CHECK: [[VBSL5_I:%.*]] = or <2 x i32> [[VBSL3_I]], [[VBSL4_I]] | ||||
| // CHECK: [[TMP5:%.*]] = bitcast <2 x i32> [[VBSL5_I]] to <2 x float> | // CHECK: [[TMP5:%.*]] = bitcast <2 x i32> [[VBSL5_I]] to <2 x float> | ||||
| // CHECK: ret <2 x float> [[TMP5]] | // CHECK: ret <2 x float> [[TMP5]] | ||||
| float32x2_t test_vbsl_f32(float32x2_t v1, float32x2_t v2, float32x2_t v3) { | float32x2_t test_vbsl_f32(uint32x2_t v1, float32x2_t v2, float32x2_t v3) { | ||||
| return vbsl_f32(v1, v2, v3); | return vbsl_f32(v1, v2, v3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vbsl_f64( | // CHECK-LABEL: @test_vbsl_f64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %v1 to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %v1 to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <1 x double> %v2 to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <1 x double> %v2 to <8 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <1 x double> %v3 to <8 x i8> | // CHECK: [[TMP2:%.*]] = bitcast <1 x double> %v3 to <8 x i8> | ||||
| // CHECK: [[VBSL1_I:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64> | // CHECK: [[VBSL1_I:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64> | ||||
| ▲ Show 20 Lines • Show All 3,575 Lines • ▼ Show 20 Lines | |||||
| // CHECK: ret <2 x i64> [[VSHL_N]] | // CHECK: ret <2 x i64> [[VSHL_N]] | ||||
| int64x2_t test_vshlq_n_s64(int64x2_t a) { | int64x2_t test_vshlq_n_s64(int64x2_t a) { | ||||
| return vshlq_n_s64(a, 3); | return vshlq_n_s64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshl_n_u8( | // CHECK-LABEL: @test_vshl_n_u8( | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <8 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSHL_N:%.*]] = shl <8 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: ret <8 x i8> [[VSHL_N]] | // CHECK: ret <8 x i8> [[VSHL_N]] | ||||
| int8x8_t test_vshl_n_u8(int8x8_t a) { | uint8x8_t test_vshl_n_u8(uint8x8_t a) { | ||||
| return vshl_n_u8(a, 3); | return vshl_n_u8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshl_n_u16( | // CHECK-LABEL: @test_vshl_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <4 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3> | // CHECK: [[VSHL_N:%.*]] = shl <4 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: ret <4 x i16> [[VSHL_N]] | // CHECK: ret <4 x i16> [[VSHL_N]] | ||||
| int16x4_t test_vshl_n_u16(int16x4_t a) { | uint16x4_t test_vshl_n_u16(uint16x4_t a) { | ||||
| return vshl_n_u16(a, 3); | return vshl_n_u16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshl_n_u32( | // CHECK-LABEL: @test_vshl_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <2 x i32> [[TMP1]], <i32 3, i32 3> | // CHECK: [[VSHL_N:%.*]] = shl <2 x i32> [[TMP1]], <i32 3, i32 3> | ||||
| // CHECK: ret <2 x i32> [[VSHL_N]] | // CHECK: ret <2 x i32> [[VSHL_N]] | ||||
| int32x2_t test_vshl_n_u32(int32x2_t a) { | uint32x2_t test_vshl_n_u32(uint32x2_t a) { | ||||
| return vshl_n_u32(a, 3); | return vshl_n_u32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshlq_n_u8( | // CHECK-LABEL: @test_vshlq_n_u8( | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSHL_N:%.*]] = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: ret <16 x i8> [[VSHL_N]] | // CHECK: ret <16 x i8> [[VSHL_N]] | ||||
| int8x16_t test_vshlq_n_u8(int8x16_t a) { | uint8x16_t test_vshlq_n_u8(uint8x16_t a) { | ||||
| return vshlq_n_u8(a, 3); | return vshlq_n_u8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshlq_n_u16( | // CHECK-LABEL: @test_vshlq_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <8 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | // CHECK: [[VSHL_N:%.*]] = shl <8 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: ret <8 x i16> [[VSHL_N]] | // CHECK: ret <8 x i16> [[VSHL_N]] | ||||
| int16x8_t test_vshlq_n_u16(int16x8_t a) { | uint16x8_t test_vshlq_n_u16(uint16x8_t a) { | ||||
| return vshlq_n_u16(a, 3); | return vshlq_n_u16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshlq_n_u32( | // CHECK-LABEL: @test_vshlq_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <4 x i32> [[TMP1]], <i32 3, i32 3, i32 3, i32 3> | // CHECK: [[VSHL_N:%.*]] = shl <4 x i32> [[TMP1]], <i32 3, i32 3, i32 3, i32 3> | ||||
| // CHECK: ret <4 x i32> [[VSHL_N]] | // CHECK: ret <4 x i32> [[VSHL_N]] | ||||
| int32x4_t test_vshlq_n_u32(int32x4_t a) { | uint32x4_t test_vshlq_n_u32(uint32x4_t a) { | ||||
| return vshlq_n_u32(a, 3); | return vshlq_n_u32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshlq_n_u64( | // CHECK-LABEL: @test_vshlq_n_u64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VSHL_N:%.*]] = shl <2 x i64> [[TMP1]], <i64 3, i64 3> | // CHECK: [[VSHL_N:%.*]] = shl <2 x i64> [[TMP1]], <i64 3, i64 3> | ||||
| // CHECK: ret <2 x i64> [[VSHL_N]] | // CHECK: ret <2 x i64> [[VSHL_N]] | ||||
| int64x2_t test_vshlq_n_u64(int64x2_t a) { | uint64x2_t test_vshlq_n_u64(uint64x2_t a) { | ||||
| return vshlq_n_u64(a, 3); | return vshlq_n_u64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshr_n_s8( | // CHECK-LABEL: @test_vshr_n_s8( | ||||
| // CHECK: [[VSHR_N:%.*]] = ashr <8 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSHR_N:%.*]] = ashr <8 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: ret <8 x i8> [[VSHR_N]] | // CHECK: ret <8 x i8> [[VSHR_N]] | ||||
| int8x8_t test_vshr_n_s8(int8x8_t a) { | int8x8_t test_vshr_n_s8(int8x8_t a) { | ||||
| return vshr_n_s8(a, 3); | return vshr_n_s8(a, 3); | ||||
| ▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines | |||||
| // CHECK: ret <2 x i64> [[VSHR_N]] | // CHECK: ret <2 x i64> [[VSHR_N]] | ||||
| int64x2_t test_vshrq_n_s64(int64x2_t a) { | int64x2_t test_vshrq_n_s64(int64x2_t a) { | ||||
| return vshrq_n_s64(a, 3); | return vshrq_n_s64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshr_n_u8( | // CHECK-LABEL: @test_vshr_n_u8( | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <8 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSHR_N:%.*]] = lshr <8 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: ret <8 x i8> [[VSHR_N]] | // CHECK: ret <8 x i8> [[VSHR_N]] | ||||
| int8x8_t test_vshr_n_u8(int8x8_t a) { | uint8x8_t test_vshr_n_u8(uint8x8_t a) { | ||||
| return vshr_n_u8(a, 3); | return vshr_n_u8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshr_n_u16( | // CHECK-LABEL: @test_vshr_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <4 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3> | // CHECK: [[VSHR_N:%.*]] = lshr <4 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: ret <4 x i16> [[VSHR_N]] | // CHECK: ret <4 x i16> [[VSHR_N]] | ||||
| int16x4_t test_vshr_n_u16(int16x4_t a) { | uint16x4_t test_vshr_n_u16(uint16x4_t a) { | ||||
| return vshr_n_u16(a, 3); | return vshr_n_u16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshr_n_u32( | // CHECK-LABEL: @test_vshr_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <2 x i32> [[TMP1]], <i32 3, i32 3> | // CHECK: [[VSHR_N:%.*]] = lshr <2 x i32> [[TMP1]], <i32 3, i32 3> | ||||
| // CHECK: ret <2 x i32> [[VSHR_N]] | // CHECK: ret <2 x i32> [[VSHR_N]] | ||||
| int32x2_t test_vshr_n_u32(int32x2_t a) { | uint32x2_t test_vshr_n_u32(uint32x2_t a) { | ||||
| return vshr_n_u32(a, 3); | return vshr_n_u32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshrq_n_u8( | // CHECK-LABEL: @test_vshrq_n_u8( | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSHR_N:%.*]] = lshr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: ret <16 x i8> [[VSHR_N]] | // CHECK: ret <16 x i8> [[VSHR_N]] | ||||
| int8x16_t test_vshrq_n_u8(int8x16_t a) { | uint8x16_t test_vshrq_n_u8(uint8x16_t a) { | ||||
| return vshrq_n_u8(a, 3); | return vshrq_n_u8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshrq_n_u16( | // CHECK-LABEL: @test_vshrq_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <8 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | // CHECK: [[VSHR_N:%.*]] = lshr <8 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: ret <8 x i16> [[VSHR_N]] | // CHECK: ret <8 x i16> [[VSHR_N]] | ||||
| int16x8_t test_vshrq_n_u16(int16x8_t a) { | uint16x8_t test_vshrq_n_u16(uint16x8_t a) { | ||||
| return vshrq_n_u16(a, 3); | return vshrq_n_u16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshrq_n_u32( | // CHECK-LABEL: @test_vshrq_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <4 x i32> [[TMP1]], <i32 3, i32 3, i32 3, i32 3> | // CHECK: [[VSHR_N:%.*]] = lshr <4 x i32> [[TMP1]], <i32 3, i32 3, i32 3, i32 3> | ||||
| // CHECK: ret <4 x i32> [[VSHR_N]] | // CHECK: ret <4 x i32> [[VSHR_N]] | ||||
| int32x4_t test_vshrq_n_u32(int32x4_t a) { | uint32x4_t test_vshrq_n_u32(uint32x4_t a) { | ||||
| return vshrq_n_u32(a, 3); | return vshrq_n_u32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshrq_n_u64( | // CHECK-LABEL: @test_vshrq_n_u64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VSHR_N:%.*]] = lshr <2 x i64> [[TMP1]], <i64 3, i64 3> | // CHECK: [[VSHR_N:%.*]] = lshr <2 x i64> [[TMP1]], <i64 3, i64 3> | ||||
| // CHECK: ret <2 x i64> [[VSHR_N]] | // CHECK: ret <2 x i64> [[VSHR_N]] | ||||
| int64x2_t test_vshrq_n_u64(int64x2_t a) { | uint64x2_t test_vshrq_n_u64(uint64x2_t a) { | ||||
| return vshrq_n_u64(a, 3); | return vshrq_n_u64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsra_n_s8( | // CHECK-LABEL: @test_vsra_n_s8( | ||||
| // CHECK: [[VSRA_N:%.*]] = ashr <8 x i8> %b, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSRA_N:%.*]] = ashr <8 x i8> %b, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VSRA_N]] | // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VSRA_N]] | ||||
| // CHECK: ret <8 x i8> [[TMP0]] | // CHECK: ret <8 x i8> [[TMP0]] | ||||
| int8x8_t test_vsra_n_s8(int8x8_t a, int8x8_t b) { | int8x8_t test_vsra_n_s8(int8x8_t a, int8x8_t b) { | ||||
| ▲ Show 20 Lines • Show All 67 Lines • ▼ Show 20 Lines | |||||
| int64x2_t test_vsraq_n_s64(int64x2_t a, int64x2_t b) { | int64x2_t test_vsraq_n_s64(int64x2_t a, int64x2_t b) { | ||||
| return vsraq_n_s64(a, b, 3); | return vsraq_n_s64(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsra_n_u8( | // CHECK-LABEL: @test_vsra_n_u8( | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <8 x i8> %b, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSRA_N:%.*]] = lshr <8 x i8> %b, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VSRA_N]] | // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VSRA_N]] | ||||
| // CHECK: ret <8 x i8> [[TMP0]] | // CHECK: ret <8 x i8> [[TMP0]] | ||||
| int8x8_t test_vsra_n_u8(int8x8_t a, int8x8_t b) { | uint8x8_t test_vsra_n_u8(uint8x8_t a, uint8x8_t b) { | ||||
| return vsra_n_u8(a, b, 3); | return vsra_n_u8(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsra_n_u16( | // CHECK-LABEL: @test_vsra_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %b to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %b to <8 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16> | // CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16> | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <4 x i16> [[TMP3]], <i16 3, i16 3, i16 3, i16 3> | // CHECK: [[VSRA_N:%.*]] = lshr <4 x i16> [[TMP3]], <i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: [[TMP4:%.*]] = add <4 x i16> [[TMP2]], [[VSRA_N]] | // CHECK: [[TMP4:%.*]] = add <4 x i16> [[TMP2]], [[VSRA_N]] | ||||
| // CHECK: ret <4 x i16> [[TMP4]] | // CHECK: ret <4 x i16> [[TMP4]] | ||||
| int16x4_t test_vsra_n_u16(int16x4_t a, int16x4_t b) { | uint16x4_t test_vsra_n_u16(uint16x4_t a, uint16x4_t b) { | ||||
| return vsra_n_u16(a, b, 3); | return vsra_n_u16(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsra_n_u32( | // CHECK-LABEL: @test_vsra_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %b to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %b to <8 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x i32> | // CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x i32> | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <2 x i32> [[TMP3]], <i32 3, i32 3> | // CHECK: [[VSRA_N:%.*]] = lshr <2 x i32> [[TMP3]], <i32 3, i32 3> | ||||
| // CHECK: [[TMP4:%.*]] = add <2 x i32> [[TMP2]], [[VSRA_N]] | // CHECK: [[TMP4:%.*]] = add <2 x i32> [[TMP2]], [[VSRA_N]] | ||||
| // CHECK: ret <2 x i32> [[TMP4]] | // CHECK: ret <2 x i32> [[TMP4]] | ||||
| int32x2_t test_vsra_n_u32(int32x2_t a, int32x2_t b) { | uint32x2_t test_vsra_n_u32(uint32x2_t a, uint32x2_t b) { | ||||
| return vsra_n_u32(a, b, 3); | return vsra_n_u32(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsraq_n_u8( | // CHECK-LABEL: @test_vsraq_n_u8( | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <16 x i8> %b, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | // CHECK: [[VSRA_N:%.*]] = lshr <16 x i8> %b, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> | ||||
| // CHECK: [[TMP0:%.*]] = add <16 x i8> %a, [[VSRA_N]] | // CHECK: [[TMP0:%.*]] = add <16 x i8> %a, [[VSRA_N]] | ||||
| // CHECK: ret <16 x i8> [[TMP0]] | // CHECK: ret <16 x i8> [[TMP0]] | ||||
| int8x16_t test_vsraq_n_u8(int8x16_t a, int8x16_t b) { | uint8x16_t test_vsraq_n_u8(uint8x16_t a, uint8x16_t b) { | ||||
| return vsraq_n_u8(a, b, 3); | return vsraq_n_u8(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsraq_n_u16( | // CHECK-LABEL: @test_vsraq_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16> | // CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16> | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <8 x i16> [[TMP3]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | // CHECK: [[VSRA_N:%.*]] = lshr <8 x i16> [[TMP3]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: [[TMP4:%.*]] = add <8 x i16> [[TMP2]], [[VSRA_N]] | // CHECK: [[TMP4:%.*]] = add <8 x i16> [[TMP2]], [[VSRA_N]] | ||||
| // CHECK: ret <8 x i16> [[TMP4]] | // CHECK: ret <8 x i16> [[TMP4]] | ||||
| int16x8_t test_vsraq_n_u16(int16x8_t a, int16x8_t b) { | uint16x8_t test_vsraq_n_u16(uint16x8_t a, uint16x8_t b) { | ||||
| return vsraq_n_u16(a, b, 3); | return vsraq_n_u16(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsraq_n_u32( | // CHECK-LABEL: @test_vsraq_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x i32> | // CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x i32> | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <4 x i32> [[TMP3]], <i32 3, i32 3, i32 3, i32 3> | // CHECK: [[VSRA_N:%.*]] = lshr <4 x i32> [[TMP3]], <i32 3, i32 3, i32 3, i32 3> | ||||
| // CHECK: [[TMP4:%.*]] = add <4 x i32> [[TMP2]], [[VSRA_N]] | // CHECK: [[TMP4:%.*]] = add <4 x i32> [[TMP2]], [[VSRA_N]] | ||||
| // CHECK: ret <4 x i32> [[TMP4]] | // CHECK: ret <4 x i32> [[TMP4]] | ||||
| int32x4_t test_vsraq_n_u32(int32x4_t a, int32x4_t b) { | uint32x4_t test_vsraq_n_u32(uint32x4_t a, uint32x4_t b) { | ||||
| return vsraq_n_u32(a, b, 3); | return vsraq_n_u32(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsraq_n_u64( | // CHECK-LABEL: @test_vsraq_n_u64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8> | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64> | // CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64> | ||||
| // CHECK: [[VSRA_N:%.*]] = lshr <2 x i64> [[TMP3]], <i64 3, i64 3> | // CHECK: [[VSRA_N:%.*]] = lshr <2 x i64> [[TMP3]], <i64 3, i64 3> | ||||
| // CHECK: [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[VSRA_N]] | // CHECK: [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[VSRA_N]] | ||||
| // CHECK: ret <2 x i64> [[TMP4]] | // CHECK: ret <2 x i64> [[TMP4]] | ||||
| int64x2_t test_vsraq_n_u64(int64x2_t a, int64x2_t b) { | uint64x2_t test_vsraq_n_u64(uint64x2_t a, uint64x2_t b) { | ||||
| return vsraq_n_u64(a, b, 3); | return vsraq_n_u64(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshr_n_s8( | // CHECK-LABEL: @test_vrshr_n_s8( | ||||
| // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.srshl.v8i8(<8 x i8> %a, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.srshl.v8i8(<8 x i8> %a, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | ||||
| // CHECK: ret <8 x i8> [[VRSHR_N]] | // CHECK: ret <8 x i8> [[VRSHR_N]] | ||||
| int8x8_t test_vrshr_n_s8(int8x8_t a) { | int8x8_t test_vrshr_n_s8(int8x8_t a) { | ||||
| return vrshr_n_s8(a, 3); | return vrshr_n_s8(a, 3); | ||||
| ▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines | |||||
| // CHECK: ret <2 x i64> [[VRSHR_N1]] | // CHECK: ret <2 x i64> [[VRSHR_N1]] | ||||
| int64x2_t test_vrshrq_n_s64(int64x2_t a) { | int64x2_t test_vrshrq_n_s64(int64x2_t a) { | ||||
| return vrshrq_n_s64(a, 3); | return vrshrq_n_s64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshr_n_u8( | // CHECK-LABEL: @test_vrshr_n_u8( | ||||
| // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.urshl.v8i8(<8 x i8> %a, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.urshl.v8i8(<8 x i8> %a, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | ||||
| // CHECK: ret <8 x i8> [[VRSHR_N]] | // CHECK: ret <8 x i8> [[VRSHR_N]] | ||||
| int8x8_t test_vrshr_n_u8(int8x8_t a) { | uint8x8_t test_vrshr_n_u8(uint8x8_t a) { | ||||
| return vrshr_n_u8(a, 3); | return vrshr_n_u8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshr_n_u16( | // CHECK-LABEL: @test_vrshr_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.urshl.v4i16(<4 x i16> [[VRSHR_N]], <4 x i16> <i16 -3, i16 -3, i16 -3, i16 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.urshl.v4i16(<4 x i16> [[VRSHR_N]], <4 x i16> <i16 -3, i16 -3, i16 -3, i16 -3>) | ||||
| // CHECK: ret <4 x i16> [[VRSHR_N1]] | // CHECK: ret <4 x i16> [[VRSHR_N1]] | ||||
| int16x4_t test_vrshr_n_u16(int16x4_t a) { | uint16x4_t test_vrshr_n_u16(uint16x4_t a) { | ||||
| return vrshr_n_u16(a, 3); | return vrshr_n_u16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshr_n_u32( | // CHECK-LABEL: @test_vrshr_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.urshl.v2i32(<2 x i32> [[VRSHR_N]], <2 x i32> <i32 -3, i32 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.urshl.v2i32(<2 x i32> [[VRSHR_N]], <2 x i32> <i32 -3, i32 -3>) | ||||
| // CHECK: ret <2 x i32> [[VRSHR_N1]] | // CHECK: ret <2 x i32> [[VRSHR_N1]] | ||||
| int32x2_t test_vrshr_n_u32(int32x2_t a) { | uint32x2_t test_vrshr_n_u32(uint32x2_t a) { | ||||
| return vrshr_n_u32(a, 3); | return vrshr_n_u32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshrq_n_u8( | // CHECK-LABEL: @test_vrshrq_n_u8( | ||||
| // CHECK: [[VRSHR_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.urshl.v16i8(<16 x i8> %a, <16 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | // CHECK: [[VRSHR_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.urshl.v16i8(<16 x i8> %a, <16 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | ||||
| // CHECK: ret <16 x i8> [[VRSHR_N]] | // CHECK: ret <16 x i8> [[VRSHR_N]] | ||||
| int8x16_t test_vrshrq_n_u8(int8x16_t a) { | uint8x16_t test_vrshrq_n_u8(uint8x16_t a) { | ||||
| return vrshrq_n_u8(a, 3); | return vrshrq_n_u8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshrq_n_u16( | // CHECK-LABEL: @test_vrshrq_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.urshl.v8i16(<8 x i16> [[VRSHR_N]], <8 x i16> <i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.urshl.v8i16(<8 x i16> [[VRSHR_N]], <8 x i16> <i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3>) | ||||
| // CHECK: ret <8 x i16> [[VRSHR_N1]] | // CHECK: ret <8 x i16> [[VRSHR_N1]] | ||||
| int16x8_t test_vrshrq_n_u16(int16x8_t a) { | uint16x8_t test_vrshrq_n_u16(uint16x8_t a) { | ||||
| return vrshrq_n_u16(a, 3); | return vrshrq_n_u16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshrq_n_u32( | // CHECK-LABEL: @test_vrshrq_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.urshl.v4i32(<4 x i32> [[VRSHR_N]], <4 x i32> <i32 -3, i32 -3, i32 -3, i32 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.urshl.v4i32(<4 x i32> [[VRSHR_N]], <4 x i32> <i32 -3, i32 -3, i32 -3, i32 -3>) | ||||
| // CHECK: ret <4 x i32> [[VRSHR_N1]] | // CHECK: ret <4 x i32> [[VRSHR_N1]] | ||||
| int32x4_t test_vrshrq_n_u32(int32x4_t a) { | uint32x4_t test_vrshrq_n_u32(uint32x4_t a) { | ||||
| return vrshrq_n_u32(a, 3); | return vrshrq_n_u32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrshrq_n_u64( | // CHECK-LABEL: @test_vrshrq_n_u64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.urshl.v2i64(<2 x i64> [[VRSHR_N]], <2 x i64> <i64 -3, i64 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.urshl.v2i64(<2 x i64> [[VRSHR_N]], <2 x i64> <i64 -3, i64 -3>) | ||||
| // CHECK: ret <2 x i64> [[VRSHR_N1]] | // CHECK: ret <2 x i64> [[VRSHR_N1]] | ||||
| int64x2_t test_vrshrq_n_u64(int64x2_t a) { | uint64x2_t test_vrshrq_n_u64(uint64x2_t a) { | ||||
| return vrshrq_n_u64(a, 3); | return vrshrq_n_u64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsra_n_s8( | // CHECK-LABEL: @test_vrsra_n_s8( | ||||
| // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.srshl.v8i8(<8 x i8> %b, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.srshl.v8i8(<8 x i8> %b, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | ||||
| // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VRSHR_N]] | // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VRSHR_N]] | ||||
| // CHECK: ret <8 x i8> [[TMP0]] | // CHECK: ret <8 x i8> [[TMP0]] | ||||
| int8x8_t test_vrsra_n_s8(int8x8_t a, int8x8_t b) { | int8x8_t test_vrsra_n_s8(int8x8_t a, int8x8_t b) { | ||||
| ▲ Show 20 Lines • Show All 67 Lines • ▼ Show 20 Lines | |||||
| int64x2_t test_vrsraq_n_s64(int64x2_t a, int64x2_t b) { | int64x2_t test_vrsraq_n_s64(int64x2_t a, int64x2_t b) { | ||||
| return vrsraq_n_s64(a, b, 3); | return vrsraq_n_s64(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsra_n_u8( | // CHECK-LABEL: @test_vrsra_n_u8( | ||||
| // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.urshl.v8i8(<8 x i8> %b, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | // CHECK: [[VRSHR_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.urshl.v8i8(<8 x i8> %b, <8 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | ||||
| // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VRSHR_N]] | // CHECK: [[TMP0:%.*]] = add <8 x i8> %a, [[VRSHR_N]] | ||||
| // CHECK: ret <8 x i8> [[TMP0]] | // CHECK: ret <8 x i8> [[TMP0]] | ||||
| int8x8_t test_vrsra_n_u8(int8x8_t a, int8x8_t b) { | uint8x8_t test_vrsra_n_u8(uint8x8_t a, uint8x8_t b) { | ||||
| return vrsra_n_u8(a, b, 3); | return vrsra_n_u8(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsra_n_u16( | // CHECK-LABEL: @test_vrsra_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %b to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %b to <8 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16> | // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.urshl.v4i16(<4 x i16> [[VRSHR_N]], <4 x i16> <i16 -3, i16 -3, i16 -3, i16 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.urshl.v4i16(<4 x i16> [[VRSHR_N]], <4 x i16> <i16 -3, i16 -3, i16 -3, i16 -3>) | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[TMP3:%.*]] = add <4 x i16> [[TMP2]], [[VRSHR_N1]] | // CHECK: [[TMP3:%.*]] = add <4 x i16> [[TMP2]], [[VRSHR_N1]] | ||||
| // CHECK: ret <4 x i16> [[TMP3]] | // CHECK: ret <4 x i16> [[TMP3]] | ||||
| int16x4_t test_vrsra_n_u16(int16x4_t a, int16x4_t b) { | uint16x4_t test_vrsra_n_u16(uint16x4_t a, uint16x4_t b) { | ||||
| return vrsra_n_u16(a, b, 3); | return vrsra_n_u16(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsra_n_u32( | // CHECK-LABEL: @test_vrsra_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %b to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %b to <8 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x i32> | // CHECK: [[VRSHR_N:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x i32> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.urshl.v2i32(<2 x i32> [[VRSHR_N]], <2 x i32> <i32 -3, i32 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.urshl.v2i32(<2 x i32> [[VRSHR_N]], <2 x i32> <i32 -3, i32 -3>) | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[TMP3:%.*]] = add <2 x i32> [[TMP2]], [[VRSHR_N1]] | // CHECK: [[TMP3:%.*]] = add <2 x i32> [[TMP2]], [[VRSHR_N1]] | ||||
| // CHECK: ret <2 x i32> [[TMP3]] | // CHECK: ret <2 x i32> [[TMP3]] | ||||
| int32x2_t test_vrsra_n_u32(int32x2_t a, int32x2_t b) { | uint32x2_t test_vrsra_n_u32(uint32x2_t a, uint32x2_t b) { | ||||
| return vrsra_n_u32(a, b, 3); | return vrsra_n_u32(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsraq_n_u8( | // CHECK-LABEL: @test_vrsraq_n_u8( | ||||
| // CHECK: [[VRSHR_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.urshl.v16i8(<16 x i8> %b, <16 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | // CHECK: [[VRSHR_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.urshl.v16i8(<16 x i8> %b, <16 x i8> <i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3, i8 -3>) | ||||
| // CHECK: [[TMP0:%.*]] = add <16 x i8> %a, [[VRSHR_N]] | // CHECK: [[TMP0:%.*]] = add <16 x i8> %a, [[VRSHR_N]] | ||||
| // CHECK: ret <16 x i8> [[TMP0]] | // CHECK: ret <16 x i8> [[TMP0]] | ||||
| int8x16_t test_vrsraq_n_u8(int8x16_t a, int8x16_t b) { | uint8x16_t test_vrsraq_n_u8(uint8x16_t a, uint8x16_t b) { | ||||
| return vrsraq_n_u8(a, b, 3); | return vrsraq_n_u8(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsraq_n_u16( | // CHECK-LABEL: @test_vrsraq_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16> | // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.urshl.v8i16(<8 x i16> [[VRSHR_N]], <8 x i16> <i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.urshl.v8i16(<8 x i16> [[VRSHR_N]], <8 x i16> <i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3, i16 -3>) | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[TMP3:%.*]] = add <8 x i16> [[TMP2]], [[VRSHR_N1]] | // CHECK: [[TMP3:%.*]] = add <8 x i16> [[TMP2]], [[VRSHR_N1]] | ||||
| // CHECK: ret <8 x i16> [[TMP3]] | // CHECK: ret <8 x i16> [[TMP3]] | ||||
| int16x8_t test_vrsraq_n_u16(int16x8_t a, int16x8_t b) { | uint16x8_t test_vrsraq_n_u16(uint16x8_t a, uint16x8_t b) { | ||||
| return vrsraq_n_u16(a, b, 3); | return vrsraq_n_u16(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsraq_n_u32( | // CHECK-LABEL: @test_vrsraq_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x i32> | // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x i32> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.urshl.v4i32(<4 x i32> [[VRSHR_N]], <4 x i32> <i32 -3, i32 -3, i32 -3, i32 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.urshl.v4i32(<4 x i32> [[VRSHR_N]], <4 x i32> <i32 -3, i32 -3, i32 -3, i32 -3>) | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[TMP3:%.*]] = add <4 x i32> [[TMP2]], [[VRSHR_N1]] | // CHECK: [[TMP3:%.*]] = add <4 x i32> [[TMP2]], [[VRSHR_N1]] | ||||
| // CHECK: ret <4 x i32> [[TMP3]] | // CHECK: ret <4 x i32> [[TMP3]] | ||||
| int32x4_t test_vrsraq_n_u32(int32x4_t a, int32x4_t b) { | uint32x4_t test_vrsraq_n_u32(uint32x4_t a, uint32x4_t b) { | ||||
| return vrsraq_n_u32(a, b, 3); | return vrsraq_n_u32(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vrsraq_n_u64( | // CHECK-LABEL: @test_vrsraq_n_u64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8> | ||||
| // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64> | // CHECK: [[VRSHR_N:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64> | ||||
| // CHECK: [[VRSHR_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.urshl.v2i64(<2 x i64> [[VRSHR_N]], <2 x i64> <i64 -3, i64 -3>) | // CHECK: [[VRSHR_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.urshl.v2i64(<2 x i64> [[VRSHR_N]], <2 x i64> <i64 -3, i64 -3>) | ||||
| // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[TMP3:%.*]] = add <2 x i64> [[TMP2]], [[VRSHR_N1]] | // CHECK: [[TMP3:%.*]] = add <2 x i64> [[TMP2]], [[VRSHR_N1]] | ||||
| // CHECK: ret <2 x i64> [[TMP3]] | // CHECK: ret <2 x i64> [[TMP3]] | ||||
| int64x2_t test_vrsraq_n_u64(int64x2_t a, int64x2_t b) { | uint64x2_t test_vrsraq_n_u64(uint64x2_t a, uint64x2_t b) { | ||||
| return vrsraq_n_u64(a, b, 3); | return vrsraq_n_u64(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsri_n_s8( | // CHECK-LABEL: @test_vsri_n_s8( | ||||
| // CHECK: [[VSRI_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.vsri.v8i8(<8 x i8> %a, <8 x i8> %b, i32 3) | // CHECK: [[VSRI_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.vsri.v8i8(<8 x i8> %a, <8 x i8> %b, i32 3) | ||||
| // CHECK: ret <8 x i8> [[VSRI_N]] | // CHECK: ret <8 x i8> [[VSRI_N]] | ||||
| int8x8_t test_vsri_n_s8(int8x8_t a, int8x8_t b) { | int8x8_t test_vsri_n_s8(int8x8_t a, int8x8_t b) { | ||||
| return vsri_n_s8(a, b, 3); | return vsri_n_s8(a, b, 3); | ||||
| ▲ Show 20 Lines • Show All 59 Lines • ▼ Show 20 Lines | |||||
| // CHECK: ret <2 x i64> [[VSRI_N2]] | // CHECK: ret <2 x i64> [[VSRI_N2]] | ||||
| int64x2_t test_vsriq_n_s64(int64x2_t a, int64x2_t b) { | int64x2_t test_vsriq_n_s64(int64x2_t a, int64x2_t b) { | ||||
| return vsriq_n_s64(a, b, 3); | return vsriq_n_s64(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsri_n_u8( | // CHECK-LABEL: @test_vsri_n_u8( | ||||
| // CHECK: [[VSRI_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.vsri.v8i8(<8 x i8> %a, <8 x i8> %b, i32 3) | // CHECK: [[VSRI_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.vsri.v8i8(<8 x i8> %a, <8 x i8> %b, i32 3) | ||||
| // CHECK: ret <8 x i8> [[VSRI_N]] | // CHECK: ret <8 x i8> [[VSRI_N]] | ||||
| int8x8_t test_vsri_n_u8(int8x8_t a, int8x8_t b) { | uint8x8_t test_vsri_n_u8(uint8x8_t a, uint8x8_t b) { | ||||
| return vsri_n_u8(a, b, 3); | return vsri_n_u8(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsri_n_u16( | // CHECK-LABEL: @test_vsri_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %b to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %b to <8 x i8> | ||||
| // CHECK: [[VSRI_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[VSRI_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[VSRI_N1:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16> | // CHECK: [[VSRI_N1:%.*]] = bitcast <8 x i8> [[TMP1]] to <4 x i16> | ||||
| // CHECK: [[VSRI_N2:%.*]] = call <4 x i16> @llvm.aarch64.neon.vsri.v4i16(<4 x i16> [[VSRI_N]], <4 x i16> [[VSRI_N1]], i32 3) | // CHECK: [[VSRI_N2:%.*]] = call <4 x i16> @llvm.aarch64.neon.vsri.v4i16(<4 x i16> [[VSRI_N]], <4 x i16> [[VSRI_N1]], i32 3) | ||||
| // CHECK: ret <4 x i16> [[VSRI_N2]] | // CHECK: ret <4 x i16> [[VSRI_N2]] | ||||
| int16x4_t test_vsri_n_u16(int16x4_t a, int16x4_t b) { | uint16x4_t test_vsri_n_u16(uint16x4_t a, uint16x4_t b) { | ||||
| return vsri_n_u16(a, b, 3); | return vsri_n_u16(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsri_n_u32( | // CHECK-LABEL: @test_vsri_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %b to <8 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %b to <8 x i8> | ||||
| // CHECK: [[VSRI_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[VSRI_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[VSRI_N1:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x i32> | // CHECK: [[VSRI_N1:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x i32> | ||||
| // CHECK: [[VSRI_N2:%.*]] = call <2 x i32> @llvm.aarch64.neon.vsri.v2i32(<2 x i32> [[VSRI_N]], <2 x i32> [[VSRI_N1]], i32 3) | // CHECK: [[VSRI_N2:%.*]] = call <2 x i32> @llvm.aarch64.neon.vsri.v2i32(<2 x i32> [[VSRI_N]], <2 x i32> [[VSRI_N1]], i32 3) | ||||
| // CHECK: ret <2 x i32> [[VSRI_N2]] | // CHECK: ret <2 x i32> [[VSRI_N2]] | ||||
| int32x2_t test_vsri_n_u32(int32x2_t a, int32x2_t b) { | uint32x2_t test_vsri_n_u32(uint32x2_t a, uint32x2_t b) { | ||||
| return vsri_n_u32(a, b, 3); | return vsri_n_u32(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsriq_n_u8( | // CHECK-LABEL: @test_vsriq_n_u8( | ||||
| // CHECK: [[VSRI_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.vsri.v16i8(<16 x i8> %a, <16 x i8> %b, i32 3) | // CHECK: [[VSRI_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.vsri.v16i8(<16 x i8> %a, <16 x i8> %b, i32 3) | ||||
| // CHECK: ret <16 x i8> [[VSRI_N]] | // CHECK: ret <16 x i8> [[VSRI_N]] | ||||
| int8x16_t test_vsriq_n_u8(int8x16_t a, int8x16_t b) { | uint8x16_t test_vsriq_n_u8(uint8x16_t a, uint8x16_t b) { | ||||
| return vsriq_n_u8(a, b, 3); | return vsriq_n_u8(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsriq_n_u16( | // CHECK-LABEL: @test_vsriq_n_u16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <8 x i16> %b to <16 x i8> | ||||
| // CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16> | // CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <8 x i16> | ||||
| // CHECK: [[VSRI_N2:%.*]] = call <8 x i16> @llvm.aarch64.neon.vsri.v8i16(<8 x i16> [[VSRI_N]], <8 x i16> [[VSRI_N1]], i32 3) | // CHECK: [[VSRI_N2:%.*]] = call <8 x i16> @llvm.aarch64.neon.vsri.v8i16(<8 x i16> [[VSRI_N]], <8 x i16> [[VSRI_N1]], i32 3) | ||||
| // CHECK: ret <8 x i16> [[VSRI_N2]] | // CHECK: ret <8 x i16> [[VSRI_N2]] | ||||
| int16x8_t test_vsriq_n_u16(int16x8_t a, int16x8_t b) { | uint16x8_t test_vsriq_n_u16(uint16x8_t a, uint16x8_t b) { | ||||
| return vsriq_n_u16(a, b, 3); | return vsriq_n_u16(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsriq_n_u32( | // CHECK-LABEL: @test_vsriq_n_u32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <4 x i32> %b to <16 x i8> | ||||
| // CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x i32> | // CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x i32> | ||||
| // CHECK: [[VSRI_N2:%.*]] = call <4 x i32> @llvm.aarch64.neon.vsri.v4i32(<4 x i32> [[VSRI_N]], <4 x i32> [[VSRI_N1]], i32 3) | // CHECK: [[VSRI_N2:%.*]] = call <4 x i32> @llvm.aarch64.neon.vsri.v4i32(<4 x i32> [[VSRI_N]], <4 x i32> [[VSRI_N1]], i32 3) | ||||
| // CHECK: ret <4 x i32> [[VSRI_N2]] | // CHECK: ret <4 x i32> [[VSRI_N2]] | ||||
| int32x4_t test_vsriq_n_u32(int32x4_t a, int32x4_t b) { | uint32x4_t test_vsriq_n_u32(uint32x4_t a, uint32x4_t b) { | ||||
| return vsriq_n_u32(a, b, 3); | return vsriq_n_u32(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsriq_n_u64( | // CHECK-LABEL: @test_vsriq_n_u64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8> | // CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8> | ||||
| // CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64> | // CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64> | ||||
| // CHECK: [[VSRI_N2:%.*]] = call <2 x i64> @llvm.aarch64.neon.vsri.v2i64(<2 x i64> [[VSRI_N]], <2 x i64> [[VSRI_N1]], i32 3) | // CHECK: [[VSRI_N2:%.*]] = call <2 x i64> @llvm.aarch64.neon.vsri.v2i64(<2 x i64> [[VSRI_N]], <2 x i64> [[VSRI_N1]], i32 3) | ||||
| // CHECK: ret <2 x i64> [[VSRI_N2]] | // CHECK: ret <2 x i64> [[VSRI_N2]] | ||||
| int64x2_t test_vsriq_n_u64(int64x2_t a, int64x2_t b) { | uint64x2_t test_vsriq_n_u64(uint64x2_t a, uint64x2_t b) { | ||||
| return vsriq_n_u64(a, b, 3); | return vsriq_n_u64(a, b, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vsri_n_p8( | // CHECK-LABEL: @test_vsri_n_p8( | ||||
| // CHECK: [[VSRI_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.vsri.v8i8(<8 x i8> %a, <8 x i8> %b, i32 3) | // CHECK: [[VSRI_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.vsri.v8i8(<8 x i8> %a, <8 x i8> %b, i32 3) | ||||
| // CHECK: ret <8 x i8> [[VSRI_N]] | // CHECK: ret <8 x i8> [[VSRI_N]] | ||||
| poly8x8_t test_vsri_n_p8(poly8x8_t a, poly8x8_t b) { | poly8x8_t test_vsri_n_p8(poly8x8_t a, poly8x8_t b) { | ||||
| return vsri_n_p8(a, b, 3); | return vsri_n_p8(a, b, 3); | ||||
| ▲ Show 20 Lines • Show All 200 Lines • ▼ Show 20 Lines | |||||
| // CHECK: ret <8 x i16> [[VSLI_N2]] | // CHECK: ret <8 x i16> [[VSLI_N2]] | ||||
| poly16x8_t test_vsliq_n_p16(poly16x8_t a, poly16x8_t b) { | poly16x8_t test_vsliq_n_p16(poly16x8_t a, poly16x8_t b) { | ||||
| return vsliq_n_p16(a, b, 15); | return vsliq_n_p16(a, b, 15); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshlu_n_s8( | // CHECK-LABEL: @test_vqshlu_n_s8( | ||||
| // CHECK: [[VQSHLU_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshlu.v8i8(<8 x i8> %a, <8 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>) | // CHECK: [[VQSHLU_N:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshlu.v8i8(<8 x i8> %a, <8 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>) | ||||
| // CHECK: ret <8 x i8> [[VQSHLU_N]] | // CHECK: ret <8 x i8> [[VQSHLU_N]] | ||||
| int8x8_t test_vqshlu_n_s8(int8x8_t a) { | uint8x8_t test_vqshlu_n_s8(int8x8_t a) { | ||||
| return vqshlu_n_s8(a, 3); | return vqshlu_n_s8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshlu_n_s16( | // CHECK-LABEL: @test_vqshlu_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8> | ||||
| // CHECK: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | // CHECK: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16> | ||||
| // CHECK: [[VQSHLU_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshlu.v4i16(<4 x i16> [[VQSHLU_N]], <4 x i16> <i16 3, i16 3, i16 3, i16 3>) | // CHECK: [[VQSHLU_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshlu.v4i16(<4 x i16> [[VQSHLU_N]], <4 x i16> <i16 3, i16 3, i16 3, i16 3>) | ||||
| // CHECK: ret <4 x i16> [[VQSHLU_N1]] | // CHECK: ret <4 x i16> [[VQSHLU_N1]] | ||||
| int16x4_t test_vqshlu_n_s16(int16x4_t a) { | uint16x4_t test_vqshlu_n_s16(int16x4_t a) { | ||||
| return vqshlu_n_s16(a, 3); | return vqshlu_n_s16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshlu_n_s32( | // CHECK-LABEL: @test_vqshlu_n_s32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8> | ||||
| // CHECK: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | // CHECK: [[VQSHLU_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32> | ||||
| // CHECK: [[VQSHLU_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqshlu.v2i32(<2 x i32> [[VQSHLU_N]], <2 x i32> <i32 3, i32 3>) | // CHECK: [[VQSHLU_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqshlu.v2i32(<2 x i32> [[VQSHLU_N]], <2 x i32> <i32 3, i32 3>) | ||||
| // CHECK: ret <2 x i32> [[VQSHLU_N1]] | // CHECK: ret <2 x i32> [[VQSHLU_N1]] | ||||
| int32x2_t test_vqshlu_n_s32(int32x2_t a) { | uint32x2_t test_vqshlu_n_s32(int32x2_t a) { | ||||
| return vqshlu_n_s32(a, 3); | return vqshlu_n_s32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshluq_n_s8( | // CHECK-LABEL: @test_vqshluq_n_s8( | ||||
| // CHECK: [[VQSHLU_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.sqshlu.v16i8(<16 x i8> %a, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>) | // CHECK: [[VQSHLU_N:%.*]] = call <16 x i8> @llvm.aarch64.neon.sqshlu.v16i8(<16 x i8> %a, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>) | ||||
| // CHECK: ret <16 x i8> [[VQSHLU_N]] | // CHECK: ret <16 x i8> [[VQSHLU_N]] | ||||
| int8x16_t test_vqshluq_n_s8(int8x16_t a) { | uint8x16_t test_vqshluq_n_s8(int8x16_t a) { | ||||
| return vqshluq_n_s8(a, 3); | return vqshluq_n_s8(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshluq_n_s16( | // CHECK-LABEL: @test_vqshluq_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VQSHLU_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.sqshlu.v8i16(<8 x i16> [[VQSHLU_N]], <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>) | // CHECK: [[VQSHLU_N1:%.*]] = call <8 x i16> @llvm.aarch64.neon.sqshlu.v8i16(<8 x i16> [[VQSHLU_N]], <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>) | ||||
| // CHECK: ret <8 x i16> [[VQSHLU_N1]] | // CHECK: ret <8 x i16> [[VQSHLU_N1]] | ||||
| int16x8_t test_vqshluq_n_s16(int16x8_t a) { | uint16x8_t test_vqshluq_n_s16(int16x8_t a) { | ||||
| return vqshluq_n_s16(a, 3); | return vqshluq_n_s16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshluq_n_s32( | // CHECK-LABEL: @test_vqshluq_n_s32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VQSHLU_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqshlu.v4i32(<4 x i32> [[VQSHLU_N]], <4 x i32> <i32 3, i32 3, i32 3, i32 3>) | // CHECK: [[VQSHLU_N1:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqshlu.v4i32(<4 x i32> [[VQSHLU_N]], <4 x i32> <i32 3, i32 3, i32 3, i32 3>) | ||||
| // CHECK: ret <4 x i32> [[VQSHLU_N1]] | // CHECK: ret <4 x i32> [[VQSHLU_N1]] | ||||
| int32x4_t test_vqshluq_n_s32(int32x4_t a) { | uint32x4_t test_vqshluq_n_s32(int32x4_t a) { | ||||
| return vqshluq_n_s32(a, 3); | return vqshluq_n_s32(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshluq_n_s64( | // CHECK-LABEL: @test_vqshluq_n_s64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[VQSHLU_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VQSHLU_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.sqshlu.v2i64(<2 x i64> [[VQSHLU_N]], <2 x i64> <i64 3, i64 3>) | // CHECK: [[VQSHLU_N1:%.*]] = call <2 x i64> @llvm.aarch64.neon.sqshlu.v2i64(<2 x i64> [[VQSHLU_N]], <2 x i64> <i64 3, i64 3>) | ||||
| // CHECK: ret <2 x i64> [[VQSHLU_N1]] | // CHECK: ret <2 x i64> [[VQSHLU_N1]] | ||||
| int64x2_t test_vqshluq_n_s64(int64x2_t a) { | uint64x2_t test_vqshluq_n_s64(int64x2_t a) { | ||||
| return vqshluq_n_s64(a, 3); | return vqshluq_n_s64(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vshrn_n_s16( | // CHECK-LABEL: @test_vshrn_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[TMP2:%.*]] = ashr <8 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | // CHECK: [[TMP2:%.*]] = ashr <8 x i16> [[TMP1]], <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> | ||||
| // CHECK: [[VSHRN_N:%.*]] = trunc <8 x i16> [[TMP2]] to <8 x i8> | // CHECK: [[VSHRN_N:%.*]] = trunc <8 x i16> [[TMP2]] to <8 x i8> | ||||
| ▲ Show 20 Lines • Show All 118 Lines • ▼ Show 20 Lines | uint32x4_t test_vshrn_high_n_u64(uint32x2_t a, uint64x2_t b) { | ||||
| return vshrn_high_n_u64(a, b, 19); | return vshrn_high_n_u64(a, b, 19); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshrun_n_s16( | // CHECK-LABEL: @test_vqshrun_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VQSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshrun.v8i8(<8 x i16> [[VQSHRUN_N]], i32 3) | // CHECK: [[VQSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshrun.v8i8(<8 x i16> [[VQSHRUN_N]], i32 3) | ||||
| // CHECK: ret <8 x i8> [[VQSHRUN_N1]] | // CHECK: ret <8 x i8> [[VQSHRUN_N1]] | ||||
| int8x8_t test_vqshrun_n_s16(int16x8_t a) { | uint8x8_t test_vqshrun_n_s16(int16x8_t a) { | ||||
| return vqshrun_n_s16(a, 3); | return vqshrun_n_s16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshrun_n_s32( | // CHECK-LABEL: @test_vqshrun_n_s32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VQSHRUN_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshrun.v4i16(<4 x i32> [[VQSHRUN_N]], i32 9) | // CHECK: [[VQSHRUN_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqshrun.v4i16(<4 x i32> [[VQSHRUN_N]], i32 9) | ||||
| // CHECK: ret <4 x i16> [[VQSHRUN_N1]] | // CHECK: ret <4 x i16> [[VQSHRUN_N1]] | ||||
| int16x4_t test_vqshrun_n_s32(int32x4_t a) { | uint16x4_t test_vqshrun_n_s32(int32x4_t a) { | ||||
| return vqshrun_n_s32(a, 9); | return vqshrun_n_s32(a, 9); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshrun_n_s64( | // CHECK-LABEL: @test_vqshrun_n_s64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VQSHRUN_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqshrun.v2i32(<2 x i64> [[VQSHRUN_N]], i32 19) | // CHECK: [[VQSHRUN_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqshrun.v2i32(<2 x i64> [[VQSHRUN_N]], i32 19) | ||||
| // CHECK: ret <2 x i32> [[VQSHRUN_N1]] | // CHECK: ret <2 x i32> [[VQSHRUN_N1]] | ||||
| int32x2_t test_vqshrun_n_s64(int64x2_t a) { | uint32x2_t test_vqshrun_n_s64(int64x2_t a) { | ||||
| return vqshrun_n_s64(a, 19); | return vqshrun_n_s64(a, 19); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqshrun_high_n_s16( | // CHECK-LABEL: @test_vqshrun_high_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8> | ||||
| // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VQSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VQSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshrun.v8i8(<8 x i16> [[VQSHRUN_N]], i32 3) | // CHECK: [[VQSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqshrun.v8i8(<8 x i16> [[VQSHRUN_N]], i32 3) | ||||
| // CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> [[VQSHRUN_N1]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15> | // CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> [[VQSHRUN_N1]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15> | ||||
| ▲ Show 20 Lines • Show All 136 Lines • ▼ Show 20 Lines | uint32x4_t test_vrshrn_high_n_u64(uint32x2_t a, uint64x2_t b) { | ||||
| return vrshrn_high_n_u64(a, b, 19); | return vrshrn_high_n_u64(a, b, 19); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqrshrun_n_s16( | // CHECK-LABEL: @test_vqrshrun_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8> | ||||
| // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VQRSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqrshrun.v8i8(<8 x i16> [[VQRSHRUN_N]], i32 3) | // CHECK: [[VQRSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqrshrun.v8i8(<8 x i16> [[VQRSHRUN_N]], i32 3) | ||||
| // CHECK: ret <8 x i8> [[VQRSHRUN_N1]] | // CHECK: ret <8 x i8> [[VQRSHRUN_N1]] | ||||
| int8x8_t test_vqrshrun_n_s16(int16x8_t a) { | uint8x8_t test_vqrshrun_n_s16(int16x8_t a) { | ||||
| return vqrshrun_n_s16(a, 3); | return vqrshrun_n_s16(a, 3); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqrshrun_n_s32( | // CHECK-LABEL: @test_vqrshrun_n_s32( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8> | ||||
| // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32> | ||||
| // CHECK: [[VQRSHRUN_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqrshrun.v4i16(<4 x i32> [[VQRSHRUN_N]], i32 9) | // CHECK: [[VQRSHRUN_N1:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqrshrun.v4i16(<4 x i32> [[VQRSHRUN_N]], i32 9) | ||||
| // CHECK: ret <4 x i16> [[VQRSHRUN_N1]] | // CHECK: ret <4 x i16> [[VQRSHRUN_N1]] | ||||
| int16x4_t test_vqrshrun_n_s32(int32x4_t a) { | uint16x4_t test_vqrshrun_n_s32(int32x4_t a) { | ||||
| return vqrshrun_n_s32(a, 9); | return vqrshrun_n_s32(a, 9); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqrshrun_n_s64( | // CHECK-LABEL: @test_vqrshrun_n_s64( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8> | ||||
| // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64> | ||||
| // CHECK: [[VQRSHRUN_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqrshrun.v2i32(<2 x i64> [[VQRSHRUN_N]], i32 19) | // CHECK: [[VQRSHRUN_N1:%.*]] = call <2 x i32> @llvm.aarch64.neon.sqrshrun.v2i32(<2 x i64> [[VQRSHRUN_N]], i32 19) | ||||
| // CHECK: ret <2 x i32> [[VQRSHRUN_N1]] | // CHECK: ret <2 x i32> [[VQRSHRUN_N1]] | ||||
| int32x2_t test_vqrshrun_n_s64(int64x2_t a) { | uint32x2_t test_vqrshrun_n_s64(int64x2_t a) { | ||||
| return vqrshrun_n_s64(a, 19); | return vqrshrun_n_s64(a, 19); | ||||
| } | } | ||||
| // CHECK-LABEL: @test_vqrshrun_high_n_s16( | // CHECK-LABEL: @test_vqrshrun_high_n_s16( | ||||
| // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8> | // CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8> | ||||
| // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | // CHECK: [[VQRSHRUN_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16> | ||||
| // CHECK: [[VQRSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqrshrun.v8i8(<8 x i16> [[VQRSHRUN_N]], i32 3) | // CHECK: [[VQRSHRUN_N1:%.*]] = call <8 x i8> @llvm.aarch64.neon.sqrshrun.v8i8(<8 x i16> [[VQRSHRUN_N]], i32 3) | ||||
| // CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> [[VQRSHRUN_N1]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15> | // CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> [[VQRSHRUN_N1]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15> | ||||
| ▲ Show 20 Lines • Show All 12,214 Lines • Show Last 20 Lines | |||||