+2017-12-15 Julia Koval <julia.koval@intel.com>
+
+ * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
+ __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
+ * config/i386/sse.md (vaesenclast_<mode>): New pattern.
+ * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
+ _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
+
2017-12-15 Julia Koval <julia.koval@intel.com>
* config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
BDESC (OPTION_MASK_ISA_VAES, CODE_FOR_vaesenc_v16qi, "__builtin_ia32_vaesenc_v16qi", IX86_BUILTIN_VAESENC16, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
BDESC (OPTION_MASK_ISA_VAES, CODE_FOR_vaesenc_v32qi, "__builtin_ia32_vaesenc_v32qi", IX86_BUILTIN_VAESENC32, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI)
BDESC (OPTION_MASK_ISA_VAES, CODE_FOR_vaesenc_v64qi, "__builtin_ia32_vaesenc_v64qi", IX86_BUILTIN_VAESENC64, UNKNOWN, (int) V64QI_FTYPE_V64QI_V64QI)
+BDESC (OPTION_MASK_ISA_VAES, CODE_FOR_vaesenclast_v16qi, "__builtin_ia32_vaesenclast_v16qi", IX86_BUILTIN_VAESENCLAST16, UNKNOWN, (int) V16QI_FTYPE_V16QI_V16QI)
+BDESC (OPTION_MASK_ISA_VAES, CODE_FOR_vaesenclast_v32qi, "__builtin_ia32_vaesenclast_v32qi", IX86_BUILTIN_VAESENCLAST32, UNKNOWN, (int) V32QI_FTYPE_V32QI_V32QI)
+BDESC (OPTION_MASK_ISA_VAES, CODE_FOR_vaesenclast_v64qi, "__builtin_ia32_vaesenclast_v64qi", IX86_BUILTIN_VAESENCLAST64, UNKNOWN, (int) V64QI_FTYPE_V64QI_V64QI)
BDESC_END (ARGS2, SPECIAL_ARGS2)
UNSPEC_VAESDEC
UNSPEC_VAESDECLAST
UNSPEC_VAESENC
+ UNSPEC_VAESENCLAST
])
(define_c_enum "unspecv" [
"TARGET_VAES"
"vaesenc\t{%2, %1, %0|%0, %1, %2}"
)
+
+(define_insn "vaesenclast_<mode>"
+ [(set (match_operand:VI1_AVX512VL_F 0 "register_operand" "=v")
+ (unspec:VI1_AVX512VL_F
+ [(match_operand:VI1_AVX512VL_F 1 "register_operand" "v")
+ (match_operand:VI1_AVX512VL_F 2 "vector_operand" "vm")]
+ UNSPEC_VAESENCLAST))]
+ "TARGET_VAES"
+ "vaesenclast\t{%2, %1, %0|%0, %1, %2}"
+)
return (__m256i)__builtin_ia32_vaesenc_v32qi ((__v32qi) __A, (__v32qi) __B);
}
+extern __inline __m256i
+__attribute__((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_aesenclast_epi128 (__m256i __A, __m256i __B)
+{
+ return (__m256i)__builtin_ia32_vaesenclast_v32qi ((__v32qi) __A,
+ (__v32qi) __B);
+}
+
#ifdef __DISABLE_VAES__
#undef __DISABLE_VAES__
#pragma GCC pop_options
return (__m512i)__builtin_ia32_vaesenc_v64qi ((__v64qi) __A, (__v64qi) __B);
}
+extern __inline __m512i
+__attribute__((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_aesenclast_epi128 (__m512i __A, __m512i __B)
+{
+ return (__m512i)__builtin_ia32_vaesenclast_v64qi ((__v64qi) __A,
+ (__v64qi) __B);
+}
+
#ifdef __DISABLE_VAESF__
#undef __DISABLE_VAESF__
#pragma GCC pop_options
return (__m128i)__builtin_ia32_vaesenc_v16qi ((__v16qi) __A, (__v16qi) __B);
}
+extern __inline __m128i
+__attribute__((__gnu_inline__, __always_inline__, __artificial__))
+_mm_aesenclast_epi128 (__m128i __A, __m128i __B)
+{
+ return (__m128i)__builtin_ia32_vaesenclast_v16qi ((__v16qi) __A,
+ (__v16qi) __B);
+}
+
#ifdef __DISABLE_VAESVL__
#undef __DISABLE_VAESVL__
#pragma GCC pop_options
+2017-12-15 Julia Koval <julia.koval@intel.com>
+
+ * gcc.target/i386/avx512f-aesenclast-2.c: New test.
+ * gcc.target/i386/avx512vl-aesenclast-2.c: Ditto.
+ * gcc.target/i386/avx512fvl-vaes-1.c: Handle new intrinsics.
+
2017-12-15 Julia Koval <julia.koval@intel.com>
* gcc.target/i386/avx512f-aesenc-2.c: New test.
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512f -mvaes" } */
+/* { dg-require-effective-target avx512f } */
+/* { dg-require-effective-target avx512vaes } */
+
+#define AVX512F
+
+#define VAES
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 32)
+
+#include "avx512f-mask-type.h"
+
+static void
+CALC (unsigned int *r)
+{
+ for (int i = 0; i < SIZE; i+=4)
+ {
+ r[i] = 0xfbcda11;
+ r[i + 1] = 0x238dd93f;
+ r[i + 2] = 0x4adc62c0;
+ r[i + 3] = 0x3efbcb88;
+ }
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_ud) res1, src1, src2;
+ MASK_TYPE mask = MASK_VALUE;
+ unsigned int res_ref[SIZE];
+
+ for (int i = 0; i < SIZE; i+=4)
+ {
+ src1.a[i] = 0x5d53475d;
+ src1.a[i + 1] = 0x63746f72;
+ src1.a[i + 2] = 0x73745665;
+ src1.a[i + 3] = 0x7b5b5465;
+ src2.a[i] = 0x726f6e5d;
+ src2.a[i + 1] = 0x5b477565;
+ src2.a[i + 2] = 0x68617929;
+ src2.a[i + 3] = 0x48692853;
+ }
+
+ CALC (res_ref);
+ res1.x = INTRINSIC (_aesenclast_epi128) (src2.x, src1.x);
+
+ if (UNION_CHECK (AVX512F_LEN, i_ud) (res1, res_ref))
+ abort ();
+}
/* { dg-final { scan-assembler-times "vaesdec\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\r]*%zmm\[0-9\]+\[^\n\r]*%zmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesdeclast\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\{\n\]*%zmm\[0-9\]+\[^\{\n\]*%zmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesenc\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\n\r]*%zmm\[0-9\]+\[^\n\r]*%zmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vaesenclast\[ \\t\]+\[^\{\n\]*%zmm\[0-9\]+\[^\{\n\]*%zmm\[0-9\]+\[^\{\n\]*%zmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesdec\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesdeclast\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesenc\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vaesenclast\[ \\t\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+\[^\{\n\]*%ymm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesdec\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesdeclast\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
/* { dg-final { scan-assembler-times "vaesenc\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
+/* { dg-final { scan-assembler-times "vaesenclast\[ \\t\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+\[^\{\n\]*%xmm\[0-9\]+(?:\n|\[ \\t\]+#)" 1 } } */
#include <immintrin.h>
x = _mm512_aesdec_epi128 (x, y);
x = _mm512_aesdeclast_epi128 (x, y);
x = _mm512_aesenc_epi128 (x, y);
+ x = _mm512_aesenclast_epi128 (x, y);
x256 = _mm256_aesdec_epi128 (x256, y256);
x256 = _mm256_aesdeclast_epi128 (x256, y256);
x256 = _mm256_aesenc_epi128 (x256, y256);
+ x256 = _mm256_aesenclast_epi128 (x256, y256);
x128 = _mm_aesdec_epi128 (x128, y128);
x128 = _mm_aesdeclast_epi128 (x128, y128);
x128 = _mm_aesenc_epi128 (x128, y128);
+ x128 = _mm_aesenclast_epi128 (x128, y128);
}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -mavx512vl -mvaes" } */
+/* { dg-require-effective-target avx512vl } */
+/* { dg-require-effective-target avx512bw } */
+/* { dg-require-effective-target avx512vaes } */
+
+#define AVX512VL
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512f-aesenclast-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512f-aesenclast-2.c"
projects / gcc.git / commitdiff