[FFmpeg-devel] [PATCH] checkasm/vp9: benchmark all sub-IDCTs (but not WHT or ADST).
Ronald S. Bultje
rsbultje at gmail.com
Tue Dec 27 17:07:29 EET 2016
Hi,
On Tue, Dec 20, 2016 at 7:10 PM, Ronald S. Bultje <rsbultje at gmail.com>
wrote:
> Hi,
>
> On Mon, Dec 5, 2016 at 4:15 PM, Ronald S. Bultje <rsbultje at gmail.com>
> wrote:
>
>> ---
>> tests/checkasm/vp9dsp.c | 22 ++++++++++++++--------
>> 1 file changed, 14 insertions(+), 8 deletions(-)
>>
>> diff --git a/tests/checkasm/vp9dsp.c b/tests/checkasm/vp9dsp.c
>> index 441041c..f32b97c 100644
>> --- a/tests/checkasm/vp9dsp.c
>> +++ b/tests/checkasm/vp9dsp.c
>> @@ -331,15 +331,20 @@ static void check_itxfm(void)
>> int n_txtps = tx < TX_32X32 ? N_TXFM_TYPES : 1;
>>
>> for (txtp = 0; txtp < n_txtps; txtp++) {
>> - if (check_func(dsp.itxfm_add[tx][txtp],
>> "vp9_inv_%s_%dx%d_add_%d",
>> - tx == 4 ? "wht_wht" : txtp_types[txtp],
>> sz, sz,
>> - bit_depth)) {
>> - randomize_buffers();
>> - ftx(coef, tx, txtp, sz, bit_depth);
>> -
>> - for (sub = (txtp == 0) ? 1 : 2; sub <= sz; sub <<=
>> 1) {
>> + // skip testing sub-IDCTs for WHT or ADST since they
>> don't
>> + // implement it in any of the SIMD functions. If they do,
>> + // consider changing this to ensure we have complete test
>> + // coverage
>> + for (sub = (txtp == 0 && tx < 4) ? 1 : sz; sub <= sz;
>> sub <<= 1) {
>> + if (check_func(dsp.itxfm_add[tx][txtp],
>> + "vp9_inv_%s_%dx%d_sub%d_add_%d",
>> + tx == 4 ? "wht_wht" :
>> txtp_types[txtp],
>> + sz, sz, sub, bit_depth)) {
>> int eob;
>>
>> + randomize_buffers();
>> + ftx(coef, tx, txtp, sz, bit_depth);
>> +
>> if (sub < sz) {
>> eob = copy_subcoefs(subcoef0, coef, tx, txtp,
>> sz, sub, bit_depth);
>> @@ -357,8 +362,9 @@ static void check_itxfm(void)
>> !iszero(subcoef0, sz * sz * SIZEOF_COEF) ||
>> !iszero(subcoef1, sz * sz * SIZEOF_COEF))
>> fail();
>> +
>> + bench_new(dst, sz * SIZEOF_PIXEL, coef, eob);
>> }
>> - bench_new(dst, sz * SIZEOF_PIXEL, coef, sz * sz);
>> }
>> }
>> }
>> --
>> 2.8.1
>
>
> Ping. I will apply this with some other outstanding patches if there's no
> comments.
>
Pushed.
Ronald
More information about the ffmpeg-devel
mailing list