[FFmpeg-devel] [PATCH]levc/hevc_cabac Optimise ff_hevc_hls_residual_coding (especially ARM)

[James Almer]

On 1/19/2016 2:05 PM, John Cox wrote:
>> On 1/19/2016 9:46 AM, John Cox wrote:
>>> +// Helper fns
>>> +#ifndef hevc_mem_bits32
>>> +static av_always_inline uint32_t hevc_mem_bits32(const void * buf, const unsigned int offset)
>>> +{
>>> +    return AV_RB32((const uint8_t *)buf + (offset >> 3)) << (offset & 7);
>>> +}
>>> +#endif
>>> +
>>> +#if AV_GCC_VERSION_AT_LEAST(3,4) && !defined(hevc_clz32)
>>> +#define hevc_clz32 hevc_clz32_builtin
>>> +static av_always_inline unsigned int hevc_clz32_builtin(const uint32_t x)
>>> +{
>>> +    // __builtin_clz says it works on ints - so adjust if int is >32 bits long
>>> +    return __builtin_clz(x) - (sizeof(int) * 8 - 32);
>>
>> Why aren't you simply using ff_clz?
> 
> Because it doesn't exist? or at least I can't find it.
> 
>>> +}
>>> +#endif
>>> +
>>> +// It is unlikely that we will ever need this but include for completeness
>>
>> There are at least two compilers we support that don't define __GNUC__, so
>> it would be used.
>> And in any case, isn't all this duplicating ff_clz, which is available in
>> libavutil/inthmath.h?
> 
> Are you sure of that?  I can find ff_ctz but no ff_clz...
> I would happily be wrong.

I assume you're writing this patch for the ffmpeg 2.8 branch or older, which you shouldn't.
Always use the master branch. You'll find ff_clz there.