This is an archive of the discontinued LLVM Phabricator instance.

[X86] Support intrinsic _mm_broadcastsi128_si256
ClosedPublic

Authored by skan on Mar 10 2020, 1:46 AM.

Reviewers

Commits

LGTM

This revision is now accepted and ready to land.Mar 10 2020, 10:55 AM

This revision was automatically updated to reflect the committed changes.

Path

Size

clang/

lib/

Headers/

2 lines

test/

CodeGen/

6 lines

	Show First 20 Lines • Show All 734 Lines • ▼ Show 20 Lines
	}			}

	static __inline__ __m256i __DEFAULT_FN_ATTRS256			static __inline__ __m256i __DEFAULT_FN_ATTRS256
	_mm256_broadcastsi128_si256(__m128i __X)			_mm256_broadcastsi128_si256(__m128i __X)
	{			{
	return (__m256i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 1, 0, 1);			return (__m256i)__builtin_shufflevector((__v2di)__X, (__v2di)__X, 0, 1, 0, 1);
	}			}

				#define _mm_broadcastsi128_si256(X) _mm256_broadcastsi128_si256(X)

	#define _mm_blend_epi32(V1, V2, M) \			#define _mm_blend_epi32(V1, V2, M) \
	(__m128i)__builtin_ia32_pblendd128((__v4si)(__m128i)(V1), \			(__m128i)__builtin_ia32_pblendd128((__v4si)(__m128i)(V1), \
	(__v4si)(__m128i)(V2), (int)(M))			(__v4si)(__m128i)(V2), (int)(M))

	#define _mm256_blend_epi32(V1, V2, M) \			#define _mm256_blend_epi32(V1, V2, M) \
	(__m256i)__builtin_ia32_pblendd256((__v8si)(__m256i)(V1), \			(__m256i)__builtin_ia32_pblendd256((__v8si)(__m256i)(V1), \
	(__v8si)(__m256i)(V2), (int)(M))			(__v8si)(__m256i)(V2), (int)(M))

	▲ Show 20 Lines • Show All 396 Lines • Show Last 20 Lines

	Show First 20 Lines • Show All 202 Lines • ▼ Show 20 Lines
	}			}

	__m256i test_mm256_broadcastsi128_si256(__m128i a) {			__m256i test_mm256_broadcastsi128_si256(__m128i a) {
	// CHECK-LABEL: test_mm256_broadcastsi128_si256			// CHECK-LABEL: test_mm256_broadcastsi128_si256
	// CHECK: shufflevector <2 x i64> %{{.}}, <2 x i64> %{{.}}, <4 x i32> <i32 0, i32 1, i32 0, i32 1>			// CHECK: shufflevector <2 x i64> %{{.}}, <2 x i64> %{{.}}, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
	return _mm256_broadcastsi128_si256(a);			return _mm256_broadcastsi128_si256(a);
	}			}

				__m256i test_mm_broadcastsi128_si256(__m128i a) {
				// CHECK-LABEL: test_mm_broadcastsi128_si256
				// CHECK: shufflevector <2 x i64> %{{.}}, <2 x i64> %{{.}}, <4 x i32> <i32 0, i32 1, i32 0, i32 1>
				return _mm_broadcastsi128_si256(a);
				}

	__m128 test_mm_broadcastss_ps(__m128 a) {			__m128 test_mm_broadcastss_ps(__m128 a) {
	// CHECK-LABEL: test_mm_broadcastss_ps			// CHECK-LABEL: test_mm_broadcastss_ps
	// CHECK-NOT: @llvm.x86.avx2.vbroadcast.ss.ps			// CHECK-NOT: @llvm.x86.avx2.vbroadcast.ss.ps
	// CHECK: shufflevector <4 x float> %{{.}}, <4 x float> %{{.}}, <4 x i32> zeroinitializer			// CHECK: shufflevector <4 x float> %{{.}}, <4 x float> %{{.}}, <4 x i32> zeroinitializer
	return _mm_broadcastss_ps(a);			return _mm_broadcastss_ps(a);
	}			}

	__m256 test_mm256_broadcastss_ps(__m128 a) {			__m256 test_mm256_broadcastss_ps(__m128 a) {
	▲ Show 20 Lines • Show All 1,068 Lines • Show Last 20 Lines