[FFmpeg-devel] checkasm/hevc_idct : update test bit depth from 8 9 and 10 to 8 10 and 12

Submitted by Yingming Fan on March 8, 2018, 7:56 a.m.

Details

Message ID 20180308075656.20434-1-fanyingming@bytedance.com
State Accepted
Commit e5b4cd4c4a9c98787b94ce33e268a61df9cf1587
Headers show

Commit Message

Yingming Fan March 8, 2018, 7:56 a.m.
From: Yingming Fan <yingmingfan@gmail.com>

---
 We have 8 10 and 12 SIMD codes, but previous checkasm hevc_idct only test 8 and 10 bit depth.
 
 tests/checkasm/hevc_idct.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Paul B Mahol March 8, 2018, 8:16 a.m.
On 3/8/18, Yingming Fan <yingmingfan@gmail.com> wrote:
> From: Yingming Fan <yingmingfan@gmail.com>
>
> ---
>  We have 8 10 and 12 SIMD codes, but previous checkasm hevc_idct only test 8
> and 10 bit depth.
>
>  tests/checkasm/hevc_idct.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/tests/checkasm/hevc_idct.c b/tests/checkasm/hevc_idct.c
> index eea712101d..c20111c2df 100644
> --- a/tests/checkasm/hevc_idct.c
> +++ b/tests/checkasm/hevc_idct.c
> @@ -87,7 +87,7 @@ void checkasm_check_hevc_idct(void)
>  {
>      int bit_depth;
>
> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>          HEVCDSPContext h;
>
>          ff_hevc_dsp_init(&h, bit_depth);
> @@ -95,7 +95,7 @@ void checkasm_check_hevc_idct(void)
>      }
>      report("idct_dc");
>
> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>          HEVCDSPContext h;
>
>          ff_hevc_dsp_init(&h, bit_depth);

Isn't this dropping 9 case?
Hendrik Leppkes March 8, 2018, 8:17 a.m.
On Thu, Mar 8, 2018 at 9:16 AM, Paul B Mahol <onemda@gmail.com> wrote:
> On 3/8/18, Yingming Fan <yingmingfan@gmail.com> wrote:
>> From: Yingming Fan <yingmingfan@gmail.com>
>>
>> ---
>>  We have 8 10 and 12 SIMD codes, but previous checkasm hevc_idct only test 8
>> and 10 bit depth.
>>
>>  tests/checkasm/hevc_idct.c | 4 ++--
>>  1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/tests/checkasm/hevc_idct.c b/tests/checkasm/hevc_idct.c
>> index eea712101d..c20111c2df 100644
>> --- a/tests/checkasm/hevc_idct.c
>> +++ b/tests/checkasm/hevc_idct.c
>> @@ -87,7 +87,7 @@ void checkasm_check_hevc_idct(void)
>>  {
>>      int bit_depth;
>>
>> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
>> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>>          HEVCDSPContext h;
>>
>>          ff_hevc_dsp_init(&h, bit_depth);
>> @@ -95,7 +95,7 @@ void checkasm_check_hevc_idct(void)
>>      }
>>      report("idct_dc");
>>
>> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
>> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>>          HEVCDSPContext h;
>>
>>          ff_hevc_dsp_init(&h, bit_depth);
>
> Isn't this dropping 9 case?

It is, but we don't have any optimizations for 9 anyway, so there is
nothing to test.

- Hendrik
Yingming Fan March 11, 2018, 2:17 a.m.
Hi, there. Is there any comment? I think i did the right fix.

Yingming Fan

> On 8 Mar 2018, at 4:17 PM, Hendrik Leppkes <h.leppkes@gmail.com> wrote:
> 
> On Thu, Mar 8, 2018 at 9:16 AM, Paul B Mahol <onemda@gmail.com> wrote:
>> On 3/8/18, Yingming Fan <yingmingfan@gmail.com> wrote:
>>> From: Yingming Fan <yingmingfan@gmail.com>
>>> 
>>> ---
>>> We have 8 10 and 12 SIMD codes, but previous checkasm hevc_idct only test 8
>>> and 10 bit depth.
>>> 
>>> tests/checkasm/hevc_idct.c | 4 ++--
>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>> 
>>> diff --git a/tests/checkasm/hevc_idct.c b/tests/checkasm/hevc_idct.c
>>> index eea712101d..c20111c2df 100644
>>> --- a/tests/checkasm/hevc_idct.c
>>> +++ b/tests/checkasm/hevc_idct.c
>>> @@ -87,7 +87,7 @@ void checkasm_check_hevc_idct(void)
>>> {
>>>     int bit_depth;
>>> 
>>> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
>>> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>>>         HEVCDSPContext h;
>>> 
>>>         ff_hevc_dsp_init(&h, bit_depth);
>>> @@ -95,7 +95,7 @@ void checkasm_check_hevc_idct(void)
>>>     }
>>>     report("idct_dc");
>>> 
>>> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
>>> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>>>         HEVCDSPContext h;
>>> 
>>>         ff_hevc_dsp_init(&h, bit_depth);
>> 
>> Isn't this dropping 9 case?
> 
> It is, but we don't have any optimizations for 9 anyway, so there is
> nothing to test.
> 
> - Hendrik
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Yingming Fan March 19, 2018, 3:38 a.m.
Hi, is there any review? This patch has been going on for a week.

Yingming Fan

> On 11 Mar 2018, at 10:17 AM, Yingming Fan <yingmingfan@gmail.com> wrote:
> 
> Hi, there. Is there any comment? I think i did the right fix.
> 
> Yingming Fan
> 
>> On 8 Mar 2018, at 4:17 PM, Hendrik Leppkes <h.leppkes@gmail.com> wrote:
>> 
>> On Thu, Mar 8, 2018 at 9:16 AM, Paul B Mahol <onemda@gmail.com> wrote:
>>> On 3/8/18, Yingming Fan <yingmingfan@gmail.com> wrote:
>>>> From: Yingming Fan <yingmingfan@gmail.com>
>>>> 
>>>> ---
>>>> We have 8 10 and 12 SIMD codes, but previous checkasm hevc_idct only test 8
>>>> and 10 bit depth.
>>>> 
>>>> tests/checkasm/hevc_idct.c | 4 ++--
>>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>>> 
>>>> diff --git a/tests/checkasm/hevc_idct.c b/tests/checkasm/hevc_idct.c
>>>> index eea712101d..c20111c2df 100644
>>>> --- a/tests/checkasm/hevc_idct.c
>>>> +++ b/tests/checkasm/hevc_idct.c
>>>> @@ -87,7 +87,7 @@ void checkasm_check_hevc_idct(void)
>>>> {
>>>>    int bit_depth;
>>>> 
>>>> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
>>>> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>>>>        HEVCDSPContext h;
>>>> 
>>>>        ff_hevc_dsp_init(&h, bit_depth);
>>>> @@ -95,7 +95,7 @@ void checkasm_check_hevc_idct(void)
>>>>    }
>>>>    report("idct_dc");
>>>> 
>>>> -    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
>>>> +    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
>>>>        HEVCDSPContext h;
>>>> 
>>>>        ff_hevc_dsp_init(&h, bit_depth);
>>> 
>>> Isn't this dropping 9 case?
>> 
>> It is, but we don't have any optimizations for 9 anyway, so there is
>> nothing to test.
>> 
>> - Hendrik
>> _______________________________________________
>> ffmpeg-devel mailing list
>> ffmpeg-devel@ffmpeg.org
>> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
James Almer March 19, 2018, 3:57 a.m.
On 3/19/2018 12:38 AM, Yingming Fan wrote:
> Hi, is there any review? This patch has been going on for a week.
> 
> Yingming Fan

Pushed, sorry for the delay.

Patch hide | download patch | download mbox

diff --git a/tests/checkasm/hevc_idct.c b/tests/checkasm/hevc_idct.c
index eea712101d..c20111c2df 100644
--- a/tests/checkasm/hevc_idct.c
+++ b/tests/checkasm/hevc_idct.c
@@ -87,7 +87,7 @@  void checkasm_check_hevc_idct(void)
 {
     int bit_depth;
 
-    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
+    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
         HEVCDSPContext h;
 
         ff_hevc_dsp_init(&h, bit_depth);
@@ -95,7 +95,7 @@  void checkasm_check_hevc_idct(void)
     }
     report("idct_dc");
 
-    for (bit_depth = 8; bit_depth <= 10; bit_depth++) {
+    for (bit_depth = 8; bit_depth <= 12; bit_depth += 2) {
         HEVCDSPContext h;
 
         ff_hevc_dsp_init(&h, bit_depth);