[FFmpeg-devel] [PATCH 02/11] avutil: add AVX-512 flags

[James Almer]

On 11/9/2017 8:58 AM, James Darnley wrote:
> ---
>  libavutil/cpu.c       | 6 +++++-
>  libavutil/cpu.h       | 1 +
>  libavutil/tests/cpu.c | 1 +
>  libavutil/x86/cpu.h   | 2 ++
>  4 files changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/libavutil/cpu.c b/libavutil/cpu.c
> index c8401b8258..6548cc3042 100644
> --- a/libavutil/cpu.c
> +++ b/libavutil/cpu.c
> @@ -80,7 +80,8 @@ void av_force_cpu_flags(int arg){
>                      AV_CPU_FLAG_XOP      |
>                      AV_CPU_FLAG_FMA3     |
>                      AV_CPU_FLAG_FMA4     |
> -                    AV_CPU_FLAG_AVX2     ))
> +                    AV_CPU_FLAG_AVX2     |
> +                    AV_CPU_FLAG_AVX512   ))
>          && !(arg & AV_CPU_FLAG_MMX)) {
>          av_log(NULL, AV_LOG_WARNING, "MMX implied by specified flags\n");
>          arg |= AV_CPU_FLAG_MMX;
> @@ -126,6 +127,7 @@ int av_parse_cpu_flags(const char *s)
>  #define CPUFLAG_AVX2     (AV_CPU_FLAG_AVX2     | CPUFLAG_AVX)
>  #define CPUFLAG_BMI2     (AV_CPU_FLAG_BMI2     | AV_CPU_FLAG_BMI1)
>  #define CPUFLAG_AESNI    (AV_CPU_FLAG_AESNI    | CPUFLAG_SSE42)
> +#define CPUFLAG_AVX512   (AV_CPU_FLAG_AVX512   | CPUFLAG_AVX2)
>      static const AVOption cpuflags_opts[] = {
>          { "flags"   , NULL, 0, AV_OPT_TYPE_FLAGS, { .i64 = 0 }, INT64_MIN, INT64_MAX, .unit = "flags" },
>  #if   ARCH_PPC
> @@ -154,6 +156,7 @@ int av_parse_cpu_flags(const char *s)
>          { "3dnowext", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPUFLAG_3DNOWEXT     },    .unit = "flags" },
>          { "cmov",     NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_CMOV     },    .unit = "flags" },
>          { "aesni"   , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPUFLAG_AESNI        },    .unit = "flags" },
> +        { "avx512"  , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = CPUFLAG_AVX512       },    .unit = "flags" },
>  #elif ARCH_ARM
>          { "armv5te",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ARMV5TE  },    .unit = "flags" },
>          { "armv6",    NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_ARMV6    },    .unit = "flags" },
> @@ -216,6 +219,7 @@ int av_parse_cpu_caps(unsigned *flags, const char *s)
>          { "3dnowext", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_3DNOWEXT },    .unit = "flags" },
>          { "cmov",     NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_CMOV     },    .unit = "flags" },
>          { "aesni",    NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AESNI    },    .unit = "flags" },
> +        { "avx512"  , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = AV_CPU_FLAG_AVX512   },    .unit = "flags" },
>  
>  #define CPU_FLAG_P2 AV_CPU_FLAG_CMOV | AV_CPU_FLAG_MMX
>  #define CPU_FLAG_P3 CPU_FLAG_P2 | AV_CPU_FLAG_MMX2 | AV_CPU_FLAG_SSE
> diff --git a/libavutil/cpu.h b/libavutil/cpu.h
> index 9e5d40affe..91523f3f5a 100644
> --- a/libavutil/cpu.h
> +++ b/libavutil/cpu.h
> @@ -55,6 +55,7 @@
>  #define AV_CPU_FLAG_FMA3        0x10000 ///< Haswell FMA3 functions
>  #define AV_CPU_FLAG_BMI1        0x20000 ///< Bit Manipulation Instruction Set 1
>  #define AV_CPU_FLAG_BMI2        0x40000 ///< Bit Manipulation Instruction Set 2
> +#define AV_CPU_FLAG_AVX512     0x100000 ///< AVX-512 functions

NIt: "AVX-512 functions: requires OS support even if YMM/ZMM registers
aren't used"

That's more in line with the AVX and AVX2 lines.

>  
>  #define AV_CPU_FLAG_ALTIVEC      0x0001 ///< standard
>  #define AV_CPU_FLAG_VSX          0x0002 ///< ISA 2.06
> diff --git a/libavutil/tests/cpu.c b/libavutil/tests/cpu.c
> index f02a54cbbb..ce45b715a0 100644
> --- a/libavutil/tests/cpu.c
> +++ b/libavutil/tests/cpu.c
> @@ -73,6 +73,7 @@ static const struct {
>      { AV_CPU_FLAG_BMI1,      "bmi1"       },
>      { AV_CPU_FLAG_BMI2,      "bmi2"       },
>      { AV_CPU_FLAG_AESNI,     "aesni"      },
> +    { AV_CPU_FLAG_AVX512,    "avx512"     },
>  #endif
>      { 0 }
>  };
> diff --git a/libavutil/x86/cpu.h b/libavutil/x86/cpu.h
> index 309b8e746c..7f4e5d08bb 100644
> --- a/libavutil/x86/cpu.h
> +++ b/libavutil/x86/cpu.h
> @@ -50,6 +50,7 @@
>  #define X86_FMA4(flags)             CPUEXT(flags, FMA4)
>  #define X86_AVX2(flags)             CPUEXT(flags, AVX2)
>  #define X86_AESNI(flags)            CPUEXT(flags, AESNI)
> +#define X86_AVX512(flags)           CPUEXT(flags, AVX512)
>  
>  #define EXTERNAL_AMD3DNOW(flags)    CPUEXT_SUFFIX(flags, _EXTERNAL, AMD3DNOW)
>  #define EXTERNAL_AMD3DNOWEXT(flags) CPUEXT_SUFFIX(flags, _EXTERNAL, AMD3DNOWEXT)
> @@ -79,6 +80,7 @@
>  #define EXTERNAL_AVX2_FAST(flags)   CPUEXT_SUFFIX_FAST2(flags, _EXTERNAL, AVX2, AVX)
>  #define EXTERNAL_AVX2_SLOW(flags)   CPUEXT_SUFFIX_SLOW2(flags, _EXTERNAL, AVX2, AVX)
>  #define EXTERNAL_AESNI(flags)       CPUEXT_SUFFIX(flags, _EXTERNAL, AESNI)
> +#define EXTERNAL_AVX512(flags)      CPUEXT_SUFFIX(flags, _EXTERNAL, AVX512)
>  
>  #define INLINE_AMD3DNOW(flags)      CPUEXT_SUFFIX(flags, _INLINE, AMD3DNOW)
>  #define INLINE_AMD3DNOWEXT(flags)   CPUEXT_SUFFIX(flags, _INLINE, AMD3DNOWEXT)
> 

Missing minor version bump and APIChanges entry.

LGTM aside from the above.