+2014-10-29 Alexander Ivchenko <alexander.ivchenko@intel.com>
+ Maxim Kuznetsov <maxim.kuznetsov@intel.com>
+ Anna Tikhonova <anna.tikhonova@intel.com>
+ Ilya Tocar <ilya.tocar@intel.com>
+ Andrey Turetskiy <andrey.turetskiy@intel.com>
+ Ilya Verbin <ilya.verbin@intel.com>
+ Kirill Yukhin <kirill.yukhin@intel.com>
+ Michael Zolotukhin <michael.v.zolotukhin@intel.com>
+
+ * config/i386/avx512bwintrin.h: Add new intrinsics.
+ * config/i386/avx512vlbwintrin.h: Ditto.
+ * config/i386/avx512vlintrin.h: Ditto.
+
2014-10-28 Dominik Vogt <vogt@linux.vnet.ibm.com>
* godump.c (precision_to_units): New helper function.
(__mmask32) __U);
}
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpeq_epu8_mask (__m512i __A, __m512i __B)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __A,
+ (__v64qi) __B, 0,
+ (__mmask64) -1);
+}
+
extern __inline __mmask64
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_cmpeq_epi8_mask (__m512i __A, __m512i __B)
(__mmask64) -1);
}
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpeq_epu8_mask (__mmask64 __U, __m512i __A, __m512i __B)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __A,
+ (__v64qi) __B, 0,
+ __U);
+}
+
extern __inline __mmask64
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_mask_cmpeq_epi8_mask (__mmask64 __U, __m512i __A, __m512i __B)
__U);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpeq_epu16_mask (__m512i __A, __m512i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __A,
+ (__v32hi) __B, 0,
+ (__mmask32) -1);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_cmpeq_epi16_mask (__m512i __A, __m512i __B)
(__mmask32) -1);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpeq_epu16_mask (__mmask32 __U, __m512i __A, __m512i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __A,
+ (__v32hi) __B, 0,
+ __U);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_mask_cmpeq_epi16_mask (__mmask32 __U, __m512i __A, __m512i __B)
__U);
}
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpgt_epu8_mask (__m512i __A, __m512i __B)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __A,
+ (__v64qi) __B, 6,
+ (__mmask64) -1);
+}
+
extern __inline __mmask64
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_cmpgt_epi8_mask (__m512i __A, __m512i __B)
(__mmask64) -1);
}
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpgt_epu8_mask (__mmask64 __U, __m512i __A, __m512i __B)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __A,
+ (__v64qi) __B, 6,
+ __U);
+}
+
extern __inline __mmask64
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_mask_cmpgt_epi8_mask (__mmask64 __U, __m512i __A, __m512i __B)
__U);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpgt_epu16_mask (__m512i __A, __m512i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __A,
+ (__v32hi) __B, 6,
+ (__mmask32) -1);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_cmpgt_epi16_mask (__m512i __A, __m512i __B)
(__mmask32) -1);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpgt_epu16_mask (__mmask32 __U, __m512i __A, __m512i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __A,
+ (__v32hi) __B, 6,
+ __U);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm512_mask_cmpgt_epi16_mask (__mmask32 __U, __m512i __A, __m512i __B)
(__mmask32) __U);
}
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpneq_epu8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 4,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmplt_epu8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 1,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpge_epu8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 5,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmple_epu8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 2,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpneq_epu16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 4,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmplt_epu16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 1,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpge_epu16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 5,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmple_epu16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 2,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpneq_epi8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 4,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmplt_epi8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 1,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpge_epi8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 5,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmple_epi8_mask (__mmask64 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 2,
+ (__mmask64) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpneq_epi16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 4,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmplt_epi16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 1,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmpge_epi16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 5,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_mask_cmple_epi16_mask (__mmask32 __M, __m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 2,
+ (__mmask32) __M);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpneq_epu8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 4,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmplt_epu8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 1,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpge_epu8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 5,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmple_epu8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_ucmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 2,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpneq_epu16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 4,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmplt_epu16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 1,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpge_epu16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 5,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmple_epu16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_ucmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 2,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpneq_epi8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 4,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmplt_epi8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 1,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpge_epi8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 5,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask64
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmple_epi8_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask64) __builtin_ia32_cmpb512_mask ((__v64qi) __X,
+ (__v64qi) __Y, 2,
+ (__mmask64) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpneq_epi16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 4,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmplt_epi16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 1,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmpge_epi16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 5,
+ (__mmask32) - 1);
+}
+
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm512_cmple_epi16_mask (__m512i __X, __m512i __Y)
+{
+ return (__mmask32) __builtin_ia32_cmpw512_mask ((__v32hi) __X,
+ (__v32hi) __Y, 2,
+ (__mmask32) - 1);
+}
+
#ifdef __OPTIMIZE__
extern __inline __m512i
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
(__mmask16) -1);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpeq_epu8_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __A,
+ (__v16qi) __B, 0,
+ (__mmask16) -1);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpeq_epu8_mask (__mmask16 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __A,
+ (__v16qi) __B, 0,
+ __U);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpeq_epi8_mask (__mmask16 __U, __m128i __A, __m128i __B)
__U);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpeq_epu8_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpb256_mask ((__v32qi) __A,
+ (__v32qi) __B, 0,
+ (__mmask32) -1);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpeq_epi8_mask (__m256i __A, __m256i __B)
(__mmask32) -1);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpeq_epu8_mask (__mmask32 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpb256_mask ((__v32qi) __A,
+ (__v32qi) __B, 0,
+ __U);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpeq_epi8_mask (__mmask32 __U, __m256i __A, __m256i __B)
__U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpeq_epu16_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __A,
+ (__v8hi) __B, 0,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpeq_epi16_mask (__m128i __A, __m128i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpeq_epu16_mask (__mmask8 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __A,
+ (__v8hi) __B, 0, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpeq_epi16_mask (__mmask8 __U, __m128i __A, __m128i __B)
(__v8hi) __B, __U);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpeq_epu16_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpw256_mask ((__v16hi) __A,
+ (__v16hi) __B, 0,
+ (__mmask16) -1);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpeq_epi16_mask (__m256i __A, __m256i __B)
(__mmask16) -1);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpeq_epu16_mask (__mmask16 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpw256_mask ((__v16hi) __A,
+ (__v16hi) __B, 0,
+ __U);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpeq_epi16_mask (__mmask16 __U, __m256i __A, __m256i __B)
__U);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpgt_epu8_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __A,
+ (__v16qi) __B, 6,
+ (__mmask16) -1);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpgt_epi8_mask (__m128i __A, __m128i __B)
(__mmask16) -1);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpgt_epu8_mask (__mmask16 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __A,
+ (__v16qi) __B, 6,
+ __U);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpgt_epi8_mask (__mmask16 __U, __m128i __A, __m128i __B)
__U);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpgt_epu8_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpb256_mask ((__v32qi) __A,
+ (__v32qi) __B, 6,
+ (__mmask32) -1);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpgt_epi8_mask (__m256i __A, __m256i __B)
(__mmask32) -1);
}
+extern __inline __mmask32
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpgt_epu8_mask (__mmask32 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask32) __builtin_ia32_ucmpb256_mask ((__v32qi) __A,
+ (__v32qi) __B, 6,
+ __U);
+}
+
extern __inline __mmask32
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpgt_epi8_mask (__mmask32 __U, __m256i __A, __m256i __B)
__U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpgt_epu16_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __A,
+ (__v8hi) __B, 6,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpgt_epi16_mask (__m128i __A, __m128i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpgt_epu16_mask (__mmask8 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __A,
+ (__v8hi) __B, 6, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpgt_epi16_mask (__mmask8 __U, __m128i __A, __m128i __B)
(__v8hi) __B, __U);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpgt_epu16_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpw256_mask ((__v16hi) __A,
+ (__v16hi) __B, 6,
+ (__mmask16) -1);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpgt_epi16_mask (__m256i __A, __m256i __B)
(__mmask16) -1);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpgt_epu16_mask (__mmask16 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask16) __builtin_ia32_ucmpw256_mask ((__v16hi) __A,
+ (__v16hi) __B, 6,
+ __U);
+}
+
extern __inline __mmask16
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpgt_epi16_mask (__mmask16 __U, __m256i __A, __m256i __B)
(__v8hi) __W, __M);
}
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epu8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 4,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epu8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 1,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epu8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 5,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epu8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_ucmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 2,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epu16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 4,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epu16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 1,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epu16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 5,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epu16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 2,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epi8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_cmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 4,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epi8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_cmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 1,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epi8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_cmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 5,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask16
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epi8_mask (__mmask16 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask16) __builtin_ia32_cmpb128_mask ((__v16qi) __X,
+ (__v16qi) __Y, 2,
+ (__mmask16) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epi16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 4,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epi16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 1,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epi16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 5,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epi16_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw128_mask ((__v8hi) __X,
+ (__v8hi) __Y, 2,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epu8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 4,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epu8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 1,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epu8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 5,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epu8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 2,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epu16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 4,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epu16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 1,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epu16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 5,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epu16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 2,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epi8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 4,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epi8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 1,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epi8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 5,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epi8_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpb256_mask ((__v32qi) __X,
+ (__v32qi) __Y, 2,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epi16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 4,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epi16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 1,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epi16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 5,
+ (__mmask8) __M);
+}
+
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epi16_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpw256_mask ((__v16hi) __X,
+ (__v16hi) __Y, 2,
+ (__mmask8) __M);
+}
+
#ifdef __DISABLE_AVX512VLBW__
#undef __DISABLE_AVX512VLBW__
#pragma GCC pop_options
(__mmask8) __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpeq_epu32_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __A,
+ (__v4si) __B, 0,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpeq_epi32_mask (__m128i __A, __m128i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpeq_epu32_mask (__mmask8 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __A,
+ (__v4si) __B, 0, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpeq_epi32_mask (__mmask8 __U, __m128i __A, __m128i __B)
(__v4si) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpeq_epu32_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __A,
+ (__v8si) __B, 0,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpeq_epi32_mask (__m256i __A, __m256i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpeq_epu32_mask (__mmask8 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __A,
+ (__v8si) __B, 0, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpeq_epi32_mask (__mmask8 __U, __m256i __A, __m256i __B)
(__v8si) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpeq_epu64_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __A,
+ (__v2di) __B, 0,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpeq_epi64_mask (__m128i __A, __m128i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpeq_epu64_mask (__mmask8 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __A,
+ (__v2di) __B, 0, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpeq_epi64_mask (__mmask8 __U, __m128i __A, __m128i __B)
(__v2di) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpeq_epu64_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __A,
+ (__v4di) __B, 0,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpeq_epi64_mask (__m256i __A, __m256i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpeq_epu64_mask (__mmask8 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __A,
+ (__v4di) __B, 0, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpeq_epi64_mask (__mmask8 __U, __m256i __A, __m256i __B)
(__v4di) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpgt_epu32_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __A,
+ (__v4si) __B, 6,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpgt_epi32_mask (__m128i __A, __m128i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpgt_epu32_mask (__mmask8 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __A,
+ (__v4si) __B, 6, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpgt_epi32_mask (__mmask8 __U, __m128i __A, __m128i __B)
(__v4si) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpgt_epu32_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __A,
+ (__v8si) __B, 6,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpgt_epi32_mask (__m256i __A, __m256i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpgt_epu32_mask (__mmask8 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __A,
+ (__v8si) __B, 6, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpgt_epi32_mask (__mmask8 __U, __m256i __A, __m256i __B)
(__v8si) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_cmpgt_epu64_mask (__m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __A,
+ (__v2di) __B, 6,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpgt_epi64_mask (__m128i __A, __m128i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpgt_epu64_mask (__mmask8 __U, __m128i __A, __m128i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __A,
+ (__v2di) __B, 6, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_mask_cmpgt_epi64_mask (__mmask8 __U, __m128i __A, __m128i __B)
(__v2di) __B, __U);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_cmpgt_epu64_mask (__m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __A,
+ (__v4di) __B, 6,
+ (__mmask8) -1);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpgt_epi64_mask (__m256i __A, __m256i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpgt_epu64_mask (__mmask8 __U, __m256i __A, __m256i __B)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __A,
+ (__v4di) __B, 6, __U);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_mask_cmpgt_epi64_mask (__mmask8 __U, __m256i __A, __m256i __B)
(__mmask8) -1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epu32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpneq_epu32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epu32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmplt_epu32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epu32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpge_epu32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epu32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmple_epu32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epu64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpneq_epu64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epu64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmplt_epu64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epu64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpge_epu64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epu64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmple_epu64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epi32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpneq_epi32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epi32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmplt_epi32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epi32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpge_epi32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epi32_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd256_mask ((__v8si) __X,
+ (__v8si) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmple_epi32_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpneq_epi64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpneq_epi64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmplt_epi64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmplt_epi64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmpge_epi64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmpge_epi64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm256_mask_cmple_epi64_mask (__mmask8 __M, __m256i __X, __m256i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq256_mask ((__v4di) __X,
+ (__v4di) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm256_cmple_epi64_mask (__m256i __X, __m256i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epu32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpneq_epu32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epu32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmplt_epu32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epu32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpge_epu32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epu32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmple_epu32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epu64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpneq_epu64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epu64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmplt_epu64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epu64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpge_epu64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epu64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_ucmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmple_epu64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epi32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpneq_epi32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epi32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmplt_epi32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epi32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpge_epi32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epi32_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpd128_mask ((__v4si) __X,
+ (__v4si) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmple_epi32_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpneq_epi64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 4,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpneq_epi64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmplt_epi64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 1,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmplt_epi64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmpge_epi64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 5,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmpge_epi64_mask (__m128i __X, __m128i __Y)
(__mmask8) - 1);
}
+extern __inline __mmask8
+ __attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
+_mm_mask_cmple_epi64_mask (__mmask8 __M, __m128i __X, __m128i __Y)
+{
+ return (__mmask8) __builtin_ia32_cmpq128_mask ((__v2di) __X,
+ (__v2di) __Y, 2,
+ (__mmask8) __M);
+}
+
extern __inline __mmask8
__attribute__ ((__gnu_inline__, __always_inline__, __artificial__))
_mm_cmple_epi64_mask (__m128i __X, __m128i __Y)
+2014-10-20 Alexander Ivchenko <alexander.ivchenko@intel.com>
+ Maxim Kuznetsov <maxim.kuznetsov@intel.com>
+ Anna Tikhonova <anna.tikhonova@intel.com>
+ Ilya Tocar <ilya.tocar@intel.com>
+ Andrey Turetskiy <andrey.turetskiy@intel.com>
+ Ilya Verbin <ilya.verbin@intel.com>
+ Kirill Yukhin <kirill.yukhin@intel.com>
+ Michael Zolotukhin <michael.v.zolotukhin@intel.com>
+
+ * gcc.target/i386/avx512bw-vpcmpequb-1.c: New.
+ * gcc.target/i386/avx512bw-vpcmpequb-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpequw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpequw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgeb-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgeb-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgeub-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgeub-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgeuw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgeuw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgew-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgew-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgtub-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgtub-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgtuw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpgtuw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpleb-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpleb-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpleub-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpleub-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpleuw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpleuw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmplew-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmplew-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltb-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltb-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltub-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltub-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltuw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltuw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpltw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpneqb-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpneqb-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpnequb-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpnequb-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpnequw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpnequw-2.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpneqw-1.c: Ditto.
+ * gcc.target/i386/avx512bw-vpcmpneqw-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpequb-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpequd-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpequd-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpequq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpequq-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpequw-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpged-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgeq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgeud-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgeuq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgtub-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgtud-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgtud-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgtuq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgtuq-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpgtuw-2.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpled-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpleq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpleud-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpleuq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpltd-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpltq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpltud-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpltuq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpneqd-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpneqq-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpnequd-1.c: Ditto.
+ * gcc.target/i386/avx512vl-vpcmpnequq-1.c: Ditto.
+
2014-10-29 Alexander Ivchenko <alexander.ivchenko@intel.com>
Maxim Kuznetsov <maxim.kuznetsov@intel.com>
Anna Tikhonova <anna.tikhonova@intel.com>
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmpeq_epu8_mask (x128, x128);
+ m32 = _mm256_cmpeq_epu8_mask (x256, x256);
+ m64 = _mm512_cmpeq_epu8_mask (x512, x512);
+ m16 = _mm_mask_cmpeq_epu8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmpeq_epu8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmpeq_epu8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned char *s1, unsigned char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] == s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+ res1 = 0;
+ res2 = 0;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpeq_epu8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpeq_epu8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmpeq_epu16_mask (x128, x128);
+ m8 = _mm_mask_cmpeq_epu16_mask (3, x128, x128);
+ m16 = _mm256_cmpeq_epu16_mask (x256, x256);
+ m16 = _mm256_mask_cmpeq_epu16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmpeq_epu16_mask (3, x512, x512);
+ m32 = _mm512_cmpeq_epu16_mask (x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned short *s1, unsigned short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] == s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpeq_epu16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpeq_epu16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmpge_epi8_mask (x128, x128);
+ m32 = _mm256_cmpge_epi8_mask (x256, x256);
+ m64 = _mm512_cmpge_epi8_mask (x512, x512);
+ m16 = _mm_mask_cmpge_epi8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmpge_epi8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmpge_epi8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, char *s1, char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] >= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpge_epi8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpge_epi8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmpge_epu8_mask (x128, x128);
+ m32 = _mm256_cmpge_epu8_mask (x256, x256);
+ m64 = _mm512_cmpge_epu8_mask (x512, x512);
+ m16 = _mm_mask_cmpge_epu8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmpge_epu8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmpge_epu8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned char *s1, unsigned char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] >= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpge_epu8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpge_epu8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmpge_epu16_mask (x128, x128);
+ m16 = _mm256_cmpge_epu16_mask (x256, x256);
+ m32 = _mm512_cmpge_epu16_mask (x512, x512);
+ m8 = _mm_mask_cmpge_epu16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmpge_epu16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmpge_epu16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned short *s1, unsigned short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] >= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpge_epu16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpge_epu16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmpge_epi16_mask (x128, x128);
+ m16 = _mm256_cmpge_epi16_mask (x256, x256);
+ m32 = _mm512_cmpge_epi16_mask (x512, x512);
+ m8 = _mm_mask_cmpge_epi16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmpge_epi16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmpge_epi16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, short *s1, short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] >= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpge_epi16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpge_epi16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmpgt_epu8_mask (x128, x128);
+ m32 = _mm256_cmpgt_epu8_mask (x256, x256);
+ m64 = _mm512_cmpgt_epu8_mask (x512, x512);
+ m16 = _mm_mask_cmpgt_epu8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmpgt_epu8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmpgt_epu8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned char *s1, unsigned char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] > s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpgt_epu8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpgt_epu8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmpgt_epu16_mask (x128, x128);
+ m16 = _mm256_cmpgt_epu16_mask (x256, x256);
+ m32 = _mm512_cmpgt_epu16_mask (x512, x512);
+ m8 = _mm_mask_cmpgt_epu16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmpgt_epu16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmpgt_epu16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned short *s1, unsigned short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] > s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpgt_epu16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpgt_epu16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmple_epi8_mask (x128, x128);
+ m32 = _mm256_cmple_epi8_mask (x256, x256);
+ m64 = _mm512_cmple_epi8_mask (x512, x512);
+ m16 = _mm_mask_cmple_epi8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmple_epi8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmple_epi8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, char *s1, char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] <= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmple_epi8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmple_epi8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmple_epu8_mask (x128, x128);
+ m32 = _mm256_cmple_epu8_mask (x256, x256);
+ m64 = _mm512_cmple_epu8_mask (x512, x512);
+ m16 = _mm_mask_cmple_epu8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmple_epu8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmple_epu8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned char *s1, unsigned char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] <= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmple_epu8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmple_epu8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmple_epu16_mask (x128, x128);
+ m16 = _mm256_cmple_epu16_mask (x256, x256);
+ m32 = _mm512_cmple_epu16_mask (x512, x512);
+ m8 = _mm_mask_cmple_epu16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmple_epu16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmple_epu16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned short *s1, unsigned short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] <= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmple_epu16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmple_epu16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmple_epi16_mask (x128, x128);
+ m16 = _mm256_cmple_epi16_mask (x256, x256);
+ m32 = _mm512_cmple_epi16_mask (x512, x512);
+ m8 = _mm_mask_cmple_epi16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmple_epi16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmple_epi16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, short *s1, short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] <= s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmple_epi16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmple_epi16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmplt_epi8_mask (x128, x128);
+ m32 = _mm256_cmplt_epi8_mask (x256, x256);
+ m64 = _mm512_cmplt_epi8_mask (x512, x512);
+ m16 = _mm_mask_cmplt_epi8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmplt_epi8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmplt_epi8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, char *s1, char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] < s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmplt_epi8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmplt_epi8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmplt_epu8_mask (x128, x128);
+ m32 = _mm256_cmplt_epu8_mask (x256, x256);
+ m64 = _mm512_cmplt_epu8_mask (x512, x512);
+ m16 = _mm_mask_cmplt_epu8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmplt_epu8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmplt_epu8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned char *s1, unsigned char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] < s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmplt_epu8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmplt_epu8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmplt_epu16_mask (x128, x128);
+ m16 = _mm256_cmplt_epu16_mask (x256, x256);
+ m32 = _mm512_cmplt_epu16_mask (x512, x512);
+ m8 = _mm_mask_cmplt_epu16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmplt_epu16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmplt_epu16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned short *s1, unsigned short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] < s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmplt_epu16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmplt_epu16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmplt_epi16_mask (x128, x128);
+ m16 = _mm256_cmplt_epi16_mask (x256, x256);
+ m32 = _mm512_cmplt_epi16_mask (x512, x512);
+ m8 = _mm_mask_cmplt_epi16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmplt_epi16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmplt_epi16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, short *s1, short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] < s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmplt_epi16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmplt_epi16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpb\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmpneq_epi8_mask (x128, x128);
+ m32 = _mm256_cmpneq_epi8_mask (x256, x256);
+ m64 = _mm512_cmpneq_epi8_mask (x512, x512);
+ m16 = _mm_mask_cmpneq_epi8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmpneq_epi8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmpneq_epi8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, char *s1, char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] != s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpneq_epi8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpneq_epi8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpub\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+volatile __mmask64 m64;
+
+void extern
+avx512bw_test (void)
+{
+ m16 = _mm_cmpneq_epu8_mask (x128, x128);
+ m32 = _mm256_cmpneq_epu8_mask (x256, x256);
+ m64 = _mm512_cmpneq_epu8_mask (x512, x512);
+ m16 = _mm_mask_cmpneq_epu8_mask (3, x128, x128);
+ m32 = _mm256_mask_cmpneq_epu8_mask (3, x256, x256);
+ m64 = _mm512_mask_cmpneq_epu8_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 8)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned char *s1, unsigned char *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] != s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_b) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpneq_epu8_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpneq_epu8_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmpneq_epu16_mask (x128, x128);
+ m16 = _mm256_cmpneq_epu16_mask (x256, x256);
+ m32 = _mm512_cmpneq_epu16_mask (x512, x512);
+ m8 = _mm_mask_cmpneq_epu16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmpneq_epu16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmpneq_epu16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, unsigned short *s1, unsigned short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] != s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpneq_epu16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpneq_epu16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512bw -mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%zmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpw\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m512i x512;
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m8;
+volatile __mmask16 m16;
+volatile __mmask32 m32;
+
+void extern
+avx512bw_test (void)
+{
+ m8 = _mm_cmpneq_epi16_mask (x128, x128);
+ m16 = _mm256_cmpneq_epi16_mask (x256, x256);
+ m32 = _mm512_cmpneq_epi16_mask (x512, x512);
+ m8 = _mm_mask_cmpneq_epi16_mask (3, x128, x128);
+ m16 = _mm256_mask_cmpneq_epi16_mask (3, x256, x256);
+ m32 = _mm512_mask_cmpneq_epi16_mask (3, x512, x512);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -DAVX512BW" } */
+/* { dg-require-effective-target avx512bw } */
+
+#include "avx512f-helper.h"
+
+#define SIZE (AVX512F_LEN / 16)
+#include "avx512f-mask-type.h"
+
+void
+CALC (MASK_TYPE *r, short *s1, short *s2)
+{
+ int i;
+ *r = 0;
+ MASK_TYPE one = 1;
+
+ for (i = 0; i < SIZE; i++)
+ if (s1[i] != s2[i])
+ *r = *r | (one << i);
+}
+
+void
+TEST (void)
+{
+ int i;
+ UNION_TYPE (AVX512F_LEN, i_w) src1, src2;
+ MASK_TYPE res_ref, res1, res2;
+ MASK_TYPE mask = MASK_VALUE;
+
+ for (i = 0; i < SIZE / 2; i++)
+ {
+ src1.a[i * 2] = i;
+ src1.a[i * 2 + 1] = i * i;
+ src2.a[i * 2] = 2 * i;
+ src2.a[i * 2 + 1] = i * i;
+ }
+
+ res1 = INTRINSIC (_cmpneq_epi16_mask) (src1.x, src2.x);
+ res2 = INTRINSIC (_mask_cmpneq_epi16_mask) (mask, src1.x, src2.x);
+
+ CALC (&res_ref, src1.a, src2.a);
+
+ if (res_ref != res1)
+ abort ();
+
+ res_ref &= mask;
+
+ if (res_ref != res2)
+ abort ();
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpequb-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpequb-2.c"
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" 1 } } */
+/* { dg-final { scan-assembler-times "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n^k\]*%k\[1-7\]\[^\{\]" 1 } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpeq_epu32_mask (x128, x128);
+ m = _mm256_cmpeq_epu32_mask (x256, x256);
+ m = _mm_mask_cmpeq_epu32_mask (3, x128, x128);
+ m = _mm256_mask_cmpeq_epu32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpequd-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpequd-2.c"
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpeq_epu64_mask (x128, x128);
+ m = _mm256_cmpeq_epu64_mask (x256, x256);
+ m = _mm_mask_cmpeq_epu64_mask (3, x128, x128);
+ m = _mm256_mask_cmpeq_epu64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpequq-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpequq-2.c"
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpequw-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpequw-2.c"
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpge_epi32_mask (x128, x128);
+ m = _mm256_cmpge_epi32_mask (x256, x256);
+ m = _mm_mask_cmpge_epi32_mask (3, x128, x128);
+ m = _mm256_mask_cmpge_epi32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpge_epi64_mask (x128, x128);
+ m = _mm256_cmpge_epi64_mask (x256, x256);
+ m = _mm_mask_cmpge_epi64_mask (3, x128, x128);
+ m = _mm256_mask_cmpge_epi64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpge_epu32_mask (x128, x128);
+ m = _mm256_cmpge_epu32_mask (x256, x256);
+ m = _mm_mask_cmpge_epu32_mask (3, x128, x128);
+ m = _mm256_mask_cmpge_epu32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpge_epu64_mask (x128, x128);
+ m = _mm256_cmpge_epu64_mask (x256, x256);
+ m = _mm_mask_cmpge_epu64_mask (3, x128, x128);
+ m = _mm256_mask_cmpge_epu64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpgtub-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpgtub-2.c"
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpgt_epu32_mask (x128, x128);
+ m = _mm256_cmpgt_epu32_mask (x256, x256);
+ m = _mm_mask_cmpgt_epu32_mask (3, x128, x128);
+ m = _mm256_mask_cmpgt_epu32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpgtud-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpgtud-2.c"
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpgt_epu64_mask (x128, x128);
+ m = _mm256_cmpgt_epu64_mask (x256, x256);
+ m = _mm_mask_cmpgt_epu64_mask (3, x128, x128);
+ m = _mm256_mask_cmpgt_epu64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpgtuq-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512f-vpcmpgtuq-2.c"
--- /dev/null
+/* { dg-do run } */
+/* { dg-options "-O2 -mavx512bw -mavx512vl -DAVX512VL" } */
+/* { dg-require-effective-target avx512vl } */
+
+#define AVX512F_LEN 256
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpgtuw-2.c"
+
+#undef AVX512F_LEN
+#undef AVX512F_LEN_HALF
+
+#define AVX512F_LEN 128
+#define AVX512F_LEN_HALF 128
+#include "avx512bw-vpcmpgtuw-2.c"
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmple_epi32_mask (x128, x128);
+ m = _mm256_cmple_epi32_mask (x256, x256);
+ m = _mm_mask_cmple_epi32_mask (3, x128, x128);
+ m = _mm256_mask_cmple_epi32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmple_epi64_mask (x128, x128);
+ m = _mm256_cmple_epi64_mask (x256, x256);
+ m = _mm_mask_cmple_epi64_mask (3, x128, x128);
+ m = _mm256_mask_cmple_epi64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmple_epu32_mask (x128, x128);
+ m = _mm256_cmple_epu32_mask (x256, x256);
+ m = _mm_mask_cmple_epu32_mask (3, x128, x128);
+ m = _mm256_mask_cmple_epu32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmple_epu64_mask (x128, x128);
+ m = _mm256_cmple_epu64_mask (x256, x256);
+ m = _mm_mask_cmple_epu64_mask (3, x128, x128);
+ m = _mm256_mask_cmple_epu64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmplt_epi32_mask (x128, x128);
+ m = _mm256_cmplt_epi32_mask (x256, x256);
+ m = _mm_mask_cmplt_epi32_mask (3, x128, x128);
+ m = _mm256_mask_cmplt_epi32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmplt_epi64_mask (x128, x128);
+ m = _mm256_cmplt_epi64_mask (x256, x256);
+ m = _mm_mask_cmplt_epi64_mask (3, x128, x128);
+ m = _mm256_mask_cmplt_epi64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmplt_epu32_mask (x128, x128);
+ m = _mm256_cmplt_epu32_mask (x256, x256);
+ m = _mm_mask_cmplt_epu32_mask (3, x128, x128);
+ m = _mm256_mask_cmplt_epu32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmplt_epu64_mask (x128, x128);
+ m = _mm256_cmplt_epu64_mask (x256, x256);
+ m = _mm_mask_cmplt_epu64_mask (3, x128, x128);
+ m = _mm256_mask_cmplt_epu64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpd\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpneq_epi32_mask (x128, x128);
+ m = _mm256_cmpneq_epi32_mask (x256, x256);
+ m = _mm_mask_cmpneq_epi32_mask (3, x128, x128);
+ m = _mm256_mask_cmpneq_epi32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpneq_epi64_mask (x128, x128);
+ m = _mm256_cmpneq_epi64_mask (x256, x256);
+ m = _mm_mask_cmpneq_epi64_mask (3, x128, x128);
+ m = _mm256_mask_cmpneq_epi64_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpud\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpneq_epu32_mask (x128, x128);
+ m = _mm256_cmpneq_epu32_mask (x256, x256);
+ m = _mm_mask_cmpneq_epu32_mask (3, x128, x128);
+ m = _mm256_mask_cmpneq_epu32_mask (3, x256, x256);
+}
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-mavx512vl -O2" } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\[^\{\]" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%ymm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+/* { dg-final { scan-assembler "vpcmpuq\[ \\t\]+\[^\n\]*%xmm\[0-9\]\[^\n\]*%k\[1-7\]\{%k\[1-7\]\}" } } */
+
+#include <immintrin.h>
+
+volatile __m256i x256;
+volatile __m128i x128;
+volatile __mmask8 m;
+
+void extern
+avx512vl_test (void)
+{
+ m = _mm_cmpneq_epu64_mask (x128, x128);
+ m = _mm256_cmpneq_epu64_mask (x256, x256);
+ m = _mm_mask_cmpneq_epu64_mask (3, x128, x128);
+ m = _mm256_mask_cmpneq_epu64_mask (3, x256, x256);
+}
projects / gcc.git / commitdiff