[FFmpeg-devel] [PATCH] ARM: NEON optimised vector_fmul

[Måns Rullgård]

Laurent Desnogues wrote:
> On Mon, Aug 25, 2008 at 5:06 AM, Mans Rullgard <mans at mansr.com> wrote:
>> ---
>>  libavcodec/armv4l/dsputil_neon.c   |    2 ++
>>  libavcodec/armv4l/dsputil_neon_s.S |   17 +++++++++++++++++
>>  2 files changed, 19 insertions(+), 0 deletions(-)
>>
> [...]
>> diff --git a/libavcodec/armv4l/dsputil_neon_s.S
>> b/libavcodec/armv4l/dsputil_neon_s.S
>> index e4b809e..d1bdba1 100644
>> --- a/libavcodec/armv4l/dsputil_neon_s.S
>> +++ b/libavcodec/armv4l/dsputil_neon_s.S
>> @@ -324,6 +324,23 @@ extern ff_float_to_int16_interleave_neon
>>         pop           {r4,r5,pc}
>>         .endfunc
>>
>> +extern ff_vector_fmul_neon
>> +        mov           r3, r0
>> +        vld1.64       {d0-d3}, [r0,:128]!
>> +        vld1.64       {d4-d7}, [r1,:128]!
>> +        dmb
>
> Shouldn't the dmb be replaced with a macro depending on
> Cortex-A8 revision?

I thought about that, and I agree it should be made optional somehow.
I can't think of a reliable way to detect it, so I guess a configure
flag will have to do.

I also have no post-r1p1 hardware to test on.

-- 
M?ns Rullg?rd
mans at mansr.com