diff mbox series

[FFmpeg-devel,25/28] avcodec/x86/mpegvideoencdsp_init: Remove bits_per_raw_sample check

Message ID GV1P250MB07373304AFA4698ECCECAFF08F1C2@GV1P250MB0737.EURP250.PROD.OUTLOOK.COM
State New
Headers show
Series [FFmpeg-devel,1/9] configure: Add missing dirac_decoder->qpeldsp dependency | expand

Checks

Context Check Description
andriy/make_x86 success Make finished
andriy/make_fate_x86 success Make fate finished

Commit Message

Andreas Rheinhardt May 6, 2024, 9:52 p.m. UTC
This check has been added in 19a0729b4cfacfd90b8ee84ab0c093ff7e397e65;
this commit added high bit-depth versions of draw_edges
and added the check to indicate that the mmx version only
supports eight bits.
Yet there is no high bit depth draw_edges function any more
since d4d186d185df98492d8935a87c5b5cf369db9748; since then,
the mmx version and the C version always behave the same,
yet said check has not been removed. This commit does so.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
---
 libavcodec/x86/mpegvideoencdsp_init.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

Comments

Michael Niedermayer May 7, 2024, 12:10 a.m. UTC | #1
On Mon, May 06, 2024 at 11:52:24PM +0200, Andreas Rheinhardt wrote:
> This check has been added in 19a0729b4cfacfd90b8ee84ab0c093ff7e397e65;
> this commit added high bit-depth versions of draw_edges
> and added the check to indicate that the mmx version only
> supports eight bits.
> Yet there is no high bit depth draw_edges function any more
> since d4d186d185df98492d8935a87c5b5cf369db9748; since then,
> the mmx version and the C version always behave the same,
> yet said check has not been removed. This commit does so.
> 
> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
> ---
>  libavcodec/x86/mpegvideoencdsp_init.c | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/libavcodec/x86/mpegvideoencdsp_init.c b/libavcodec/x86/mpegvideoencdsp_init.c
> index ec174b15aa..9fa7ee4824 100644
> --- a/libavcodec/x86/mpegvideoencdsp_init.c
> +++ b/libavcodec/x86/mpegvideoencdsp_init.c
> @@ -233,9 +233,7 @@ av_cold void ff_mpegvideoencdsp_init_x86(MpegvideoEncDSPContext *c,
>          }
>          c->add_8x8basis = add_8x8basis_mmx;
>  
> -        if (avctx->bits_per_raw_sample <= 8) {
> -            c->draw_edges = draw_edges_mmx;
> -        }
> +        c->draw_edges = draw_edges_mmx;

does draw_edges_mmx support >8 bit ?
if not the code prior to the patch looks correct to me.
and the patch removes a part of >8bit support
it would be better to finish the >8bit support

thx

[...]
Andreas Rheinhardt May 7, 2024, 8:49 a.m. UTC | #2
Michael Niedermayer:
> On Mon, May 06, 2024 at 11:52:24PM +0200, Andreas Rheinhardt wrote:
>> This check has been added in 19a0729b4cfacfd90b8ee84ab0c093ff7e397e65;
>> this commit added high bit-depth versions of draw_edges
>> and added the check to indicate that the mmx version only
>> supports eight bits.
>> Yet there is no high bit depth draw_edges function any more
>> since d4d186d185df98492d8935a87c5b5cf369db9748; since then,
>> the mmx version and the C version always behave the same,
>> yet said check has not been removed. This commit does so.
>>
>> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
>> ---
>>  libavcodec/x86/mpegvideoencdsp_init.c | 4 +---
>>  1 file changed, 1 insertion(+), 3 deletions(-)
>>
>> diff --git a/libavcodec/x86/mpegvideoencdsp_init.c b/libavcodec/x86/mpegvideoencdsp_init.c
>> index ec174b15aa..9fa7ee4824 100644
>> --- a/libavcodec/x86/mpegvideoencdsp_init.c
>> +++ b/libavcodec/x86/mpegvideoencdsp_init.c
>> @@ -233,9 +233,7 @@ av_cold void ff_mpegvideoencdsp_init_x86(MpegvideoEncDSPContext *c,
>>          }
>>          c->add_8x8basis = add_8x8basis_mmx;
>>  
>> -        if (avctx->bits_per_raw_sample <= 8) {
>> -            c->draw_edges = draw_edges_mmx;
>> -        }
>> +        c->draw_edges = draw_edges_mmx;
> 
> does draw_edges_mmx support >8 bit ?
> if not the code prior to the patch looks correct to me.
> and the patch removes a part of >8bit support
> it would be better to finish the >8bit support
> 

Neither the C version nor the mmx version supports >8 bit (the former
does not check for bits per raw sample at all). And none of the users
want a high bit version. As has been said: It has been removed in
d4d186d185df98492d8935a87c5b5cf369db9748.
(If desired, a high bit depth version should IMO be a separate function
pointer in the ctx.)

- Andreas
diff mbox series

Patch

diff --git a/libavcodec/x86/mpegvideoencdsp_init.c b/libavcodec/x86/mpegvideoencdsp_init.c
index ec174b15aa..9fa7ee4824 100644
--- a/libavcodec/x86/mpegvideoencdsp_init.c
+++ b/libavcodec/x86/mpegvideoencdsp_init.c
@@ -233,9 +233,7 @@  av_cold void ff_mpegvideoencdsp_init_x86(MpegvideoEncDSPContext *c,
         }
         c->add_8x8basis = add_8x8basis_mmx;
 
-        if (avctx->bits_per_raw_sample <= 8) {
-            c->draw_edges = draw_edges_mmx;
-        }
+        c->draw_edges = draw_edges_mmx;
     }
 
     if (INLINE_AMD3DNOW(cpu_flags)) {