On 06/12/16 22:26, Anuj Phogat wrote:
> On Tue, Dec 6, 2016 at 10:58 AM, Alejandro Piñeiro
> wrote:
>> The FIXME suggest that the check should be removed.
>>
> Only if we see any performance or feature benefits in doing that.
> Last I checked I didn't see any performance
On Wed, Dec 7, 2016 at 3:38 AM, Alejandro Piñeiro wrote:
>
>
> On 07/12/16 09:36, Alejandro Piñeiro wrote:
>> On 06/12/16 22:26, Anuj Phogat wrote:
>>> On Tue, Dec 6, 2016 at 10:58 AM, Alejandro Piñeiro
>>> wrote:
The FIXME suggest that the check
On 07/12/16 09:36, Alejandro Piñeiro wrote:
> On 06/12/16 22:26, Anuj Phogat wrote:
>> On Tue, Dec 6, 2016 at 10:58 AM, Alejandro Piñeiro
>> wrote:
>>> The FIXME suggest that the check should be removed.
>>>
>> Only if we see any performance or feature benefits in doing
On 06/12/16 22:26, Anuj Phogat wrote:
> On Tue, Dec 6, 2016 at 10:58 AM, Alejandro Piñeiro
> wrote:
>> The FIXME suggest that the check should be removed.
>>
> Only if we see any performance or feature benefits in doing that.
> Last I checked I didn't see any performance
On Tue, Dec 6, 2016 at 4:26 PM, Anuj Phogat wrote:
> On Tue, Dec 6, 2016 at 10:58 AM, Alejandro Piñeiro
> wrote:
>> The FIXME suggest that the check should be removed.
>>
> Only if we see any performance or feature benefits in doing that.
> Last I
On Tue, Dec 6, 2016 at 10:58 AM, Alejandro Piñeiro wrote:
> The FIXME suggest that the check should be removed.
>
Only if we see any performance or feature benefits in doing that.
Last I checked I didn't see any performance benefits on Skylake.
I also couldn't figure out the
The FIXME suggest that the check should be removed.
This change helps the following test:
GL45-CTS.texture_cube_map_array.color_depth_attachments
to pass consistently on Skylake GT3e. Without this patch, on
Skylake GT3e that test has a ~76% pass rate, with some random
intel_do_flush_locked