[FFmpeg-devel] [PATCH] vf_psnr: sse2 optimizations for sum-squared-error.

James Almer jamrial at gmail.com
Mon Jul 13 22:14:09 CEST 2015


On 13/07/15 5:07 PM, Ronald S. Bultje wrote:
> Hi,
> 
> On Mon, Jul 13, 2015 at 3:50 PM, James Darnley <james.darnley at gmail.com>
> wrote:
> 
>> On 2015-07-13 01:34, Ronald S. Bultje wrote:
>>> Hi,
>>>
>>> On Sun, Jul 12, 2015 at 5:54 PM, Paul B Mahol <onemda at gmail.com> wrote:
>>>
>>>> On 7/12/15, Ronald S. Bultje <rsbultje at gmail.com> wrote:
>>>>> +typedef struct PSNRDSPContext {
>>>>> +    uint64_t (*sse_line)(const uint8_t *buf, const uint8_t *ref, int
>> w);
>>>>
>>>> Besides naming of function sse_line rest LGTM.
>>>>
>>>
>>> That was sort of inspired by other functions in lavfi. I'm open to
>>> alternative suggestions.
>>
>> If it was called sse after Intel's Streaming SIMD Extensions then it
>> would be "bad" because it implies it must use those instructions.  (Or
>> some reason like that.)
>>
>> If it was called sse after Sum of Squared Errors then it would be "good"
>> because the name is correctly reflecting what it does.
>>
>> Perhaps ssd is more acceptable after Sum of Squared Differences?
> 
> 
> Libavcodec calls this thing sum-of-squared-errors already (grep -i squared
> libavcodec/x86/*.asm).
> 
> Ronald

Yeah, me_cmp uses sse as abbreviation for Sum of Squared Errors, so IMO it's
fine here as well.


More information about the ffmpeg-devel mailing list