Author: jina.nahias Date: Tue Dec 5 07:42:47 2017 New Revision: 319777 URL: http://llvm.org/viewvc/llvm-project?rev=319777&view=rev Log: [x86][AVX512] Lowering kunpack intrinsics to LLVM IR
This patch, together with a matching llvm patch (https://reviews.llvm.org/D39720), implements the lowering of X86 kunpack intrinsics to IR. Differential Revision: https://reviews.llvm.org/D39719 Change-Id: Id5d3cb394ad33b98be79a6783d1d15569e2b798d Modified: cfe/trunk/lib/Headers/avx512bwintrin.h cfe/trunk/lib/Headers/avx512fintrin.h cfe/trunk/test/CodeGen/avx512bw-builtins.c cfe/trunk/test/CodeGen/avx512f-builtins.c Modified: cfe/trunk/lib/Headers/avx512bwintrin.h URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx512bwintrin.h?rev=319777&r1=319776&r2=319777&view=diff ============================================================================== --- cfe/trunk/lib/Headers/avx512bwintrin.h (original) +++ cfe/trunk/lib/Headers/avx512bwintrin.h Tue Dec 5 07:42:47 2017 @@ -1854,15 +1854,13 @@ _mm512_maskz_set1_epi8 (__mmask64 __M, c static __inline__ __mmask64 __DEFAULT_FN_ATTRS _mm512_kunpackd (__mmask64 __A, __mmask64 __B) { - return (__mmask64) __builtin_ia32_kunpckdi ((__mmask64) __A, - (__mmask64) __B); + return (__mmask64) (( __A & 0xFFFFFFFF) | ( __B << 32)); } static __inline__ __mmask32 __DEFAULT_FN_ATTRS _mm512_kunpackw (__mmask32 __A, __mmask32 __B) { - return (__mmask32) __builtin_ia32_kunpcksi ((__mmask32) __A, - (__mmask32) __B); +return (__mmask32) (( __A & 0xFFFF) | ( __B << 16)); } static __inline__ __m512i __DEFAULT_FN_ATTRS Modified: cfe/trunk/lib/Headers/avx512fintrin.h URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx512fintrin.h?rev=319777&r1=319776&r2=319777&view=diff ============================================================================== --- cfe/trunk/lib/Headers/avx512fintrin.h (original) +++ cfe/trunk/lib/Headers/avx512fintrin.h Tue Dec 5 07:42:47 2017 @@ -8787,7 +8787,7 @@ _mm512_kortestz (__mmask16 __A, __mmask1 static __inline__ __mmask16 __DEFAULT_FN_ATTRS _mm512_kunpackb (__mmask16 __A, __mmask16 __B) { - return (__mmask16) __builtin_ia32_kunpckhi ((__mmask16) __A, (__mmask16) __B); + return (__mmask16) (( __A & 0xFF) | ( __B << 8)); } static __inline__ __mmask16 __DEFAULT_FN_ATTRS Modified: cfe/trunk/test/CodeGen/avx512bw-builtins.c URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx512bw-builtins.c?rev=319777&r1=319776&r2=319777&view=diff ============================================================================== --- cfe/trunk/test/CodeGen/avx512bw-builtins.c (original) +++ cfe/trunk/test/CodeGen/avx512bw-builtins.c Tue Dec 5 07:42:47 2017 @@ -1626,16 +1626,26 @@ __m512i test_mm512_maskz_set1_epi8(__mma return _mm512_maskz_set1_epi8(__M, __A); } -__mmask64 test_mm512_kunpackd(__mmask64 __A, __mmask64 __B) { +__mmask64 test_mm512_kunpackd(__m512i __A, __m512i __B, __m512i __C, __m512i __D, __m512i __E, __m512i __F) { // CHECK-LABEL: @test_mm512_kunpackd - // CHECK: @llvm.x86.avx512.kunpck.dq - return _mm512_kunpackd(__A, __B); + // CHECK: bitcast <64 x i1> %{{.*}} to i64 + // CHECK: bitcast <64 x i1> %{{.*}} to i64 + // CHECK: and i64 %{{.*}}, 4294967295 + // CHECK: shl i64 %{{.*}}, 32 + // CHECK: or i64 %{{.*}}, %{{.*}} + // CHECK: bitcast i64 %{{.*}} to <64 x i1> + return _mm512_mask_cmpneq_epu8_mask(_mm512_kunpackd(_mm512_cmpneq_epu8_mask(__B, __A),_mm512_cmpneq_epu8_mask(__C, __D)), __E, __F); } -__mmask32 test_mm512_kunpackw(__mmask32 __A, __mmask32 __B) { +__mmask32 test_mm512_kunpackw(__m512i __A, __m512i __B, __m512i __C, __m512i __D, __m512i __E, __m512i __F) { // CHECK-LABEL: @test_mm512_kunpackw - // CHECK: @llvm.x86.avx512.kunpck.wd - return _mm512_kunpackw(__A, __B); + // CHECK: bitcast <32 x i1> %{{.*}} to i32 + // CHECK: bitcast <32 x i1> %{{.*}} to i32 + // CHECK: and i32 %{{.*}}, 65535 + // CHECK: shl i32 %{{.*}}, 16 + // CHECK: or i32 %{{.*}}, %{{.*}} + // CHECK: bitcast i32 %{{.*}} to <32 x i1> + return _mm512_mask_cmpneq_epu16_mask(_mm512_kunpackw(_mm512_cmpneq_epu16_mask(__B, __A),_mm512_cmpneq_epu16_mask(__C, __D)), __E, __F); } __m512i test_mm512_mask_loadu_epi16(__m512i __W, __mmask32 __U, void const *__P) { Modified: cfe/trunk/test/CodeGen/avx512f-builtins.c URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx512f-builtins.c?rev=319777&r1=319776&r2=319777&view=diff ============================================================================== --- cfe/trunk/test/CodeGen/avx512f-builtins.c (original) +++ cfe/trunk/test/CodeGen/avx512f-builtins.c Tue Dec 5 07:42:47 2017 @@ -6241,10 +6241,17 @@ int test_mm512_kortestz(__mmask16 __A, _ return _mm512_kortestz(__A, __B); } -__mmask16 test_mm512_kunpackb(__mmask16 __A, __mmask16 __B) { +__mmask16 test_mm512_kunpackb(__m512i __A, __m512i __B, __m512i __C, __m512i __D, __m512i __E, __m512i __F) { // CHECK-LABEL: @test_mm512_kunpackb - // CHECK: @llvm.x86.avx512.kunpck.bw - return _mm512_kunpackb(__A, __B); + // CHECK: bitcast <16 x i1> %{{.*}} to i16 + // CHECK: bitcast <16 x i1> %{{.*}} to i16 + // CHECK: and i32 %{{.*}}, 255 + // CHECK: shl i32 %{{.*}}, 8 + // CHECK: or i32 %{{.*}}, %{{.*}} + // CHECK: bitcast i16 %{{.*}} to <16 x i1> + return _mm512_mask_cmpneq_epu32_mask(_mm512_kunpackb(_mm512_cmpneq_epu32_mask(__A, __B), + _mm512_cmpneq_epu32_mask(__C, __D)), + __E, __F); } __mmask16 test_mm512_kxnor(__mmask16 __A, __mmask16 __B) { _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
