Janne Grunau <[email protected]> writes:

> On 2011-12-10 15:43:55 +0000, Mans Rullgard wrote:
>> Signed-off-by: Mans Rullgard <[email protected]>
>> ---
>>  libavcodec/arm/asm.S               |    6 +++
>>  libavcodec/arm/rv40dsp_init_neon.c |   10 ++++
>>  libavcodec/arm/rv40dsp_neon.S      |   84 
>> ++++++++++++++++++++++++++++++++++++
>>  3 files changed, 100 insertions(+), 0 deletions(-)
>> 
>> diff --git a/libavcodec/arm/asm.S b/libavcodec/arm/asm.S
>> index a124918..d9f9051 100644
>> --- a/libavcodec/arm/asm.S
>> +++ b/libavcodec/arm/asm.S
>> @@ -113,6 +113,12 @@ T       add             \rn, \rn, \rm
>>  T       ldr             \rt, [\rn]
>>  .endm
>>  
>> +.macro  ldr_dpre        rt,  rn,  rm:vararg
>> +A       ldr             \rt, [\rn, -\rm]!
>> +T       add             \rn, \rn, \rm
>
> sub?

I should probably test these things...

>> +function ff_rv40_h_loop_filter_strength_neon, export=1
>> +        pkhbt           r2,  r3,  r2,  lsl #18
>> +
>> +        ldr             r3,  [r0]
>> +        ldr_dpre        r12, r0,  r1
>> +        teq             r3,  r12
>> +        beq             1f
>> +
>> +        sub             r0,  r0,  r1,  lsl #1
>> +
>> +        vld1.32         {d4[]},   [r0,:32], r1  @ -3
>> +        vld1.32         {d0[]},   [r0,:32], r1  @ -2
>> +        vld1.32         {d4[1]},  [r0,:32], r1  @ -1
>> +        vld1.32         {d5[]},   [r0,:32], r1  @  0
>> +        vld1.32         {d1[]},   [r0,:32], r1  @  1
>> +        vld1.32         {d5[0]},  [r0,:32], r1  @  2
>> +
>> +        vpaddl.u8       q8,  q0                 @ -2, -2, -2, -2,  1,  1,  
>> 1,  1
>> +        vpaddl.u8       q9,  q2                 @ -1, -1, -3, -3,  0,  0,  
>> 2,  2
>
> comment is wrong, it's @ -1, -1, -3, -3,  2,  2,  0,  0

It's actually -3, -3, -1, -1,  2,  2,  0,  0.  The comment further down
is correct, however.

>> +        vdup.32         d30, r2                 @ beta2, beta << 2
>> +        vpadd.u16       d16, d16, d17           @ -2, -2,  1,  1
>> +        vpadd.u16       d18, d18, d19           @ -3, -1,  2,  0
>> +        vabd.u16        d16, d18, d16
>> +        vclt.u16        d16, d16, d30

-- 
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to