[Mesa-dev] [PATCH 00/18] Gallium blitter optimizations

Marek Olšák maraeo at gmail.com
Mon Aug 21 13:21:55 UTC 2017


I'll push your patch when I push my series. I guess that's all, right?

Thanks,
Marek

On Mon, Aug 21, 2017 at 3:17 AM, Dave Airlie <airlied at gmail.com> wrote:
> On 21 August 2017 at 10:58, Marek Olšák <maraeo at gmail.com> wrote:
>> On Mon, Aug 21, 2017 at 2:41 AM, Dave Airlie <airlied at gmail.com> wrote:
>>> On 21 August 2017 at 10:22, Dave Airlie <airlied at gmail.com> wrote:
>>>> Hi Gert,
>>>>
>>>> Can you test this along with the fetch shader patch Marek sent?
>>>>
>>>> I'm giving it a piglit run now.
>>>
>>> Actually that patch is probably not necessary,
>>>
>>> I think you need to fill in 0 for the 4th 2D coordinate for LD to be used.
>>>
>>> TGSI spec for LD says it takes level in the last channel of the coord, and you
>>> never set it, whereas the old blitter path set it correctly.
>>
>> radeonsi always uses TXF_LZ with u_blitter. That way level=0 is
>> implied by the instruction. The CAP demands that both TXF_LZ and
>> TEX_LZ are supported.
>
> Do we add support for the cap in r600 and make the TXF_LZ CAP
> mandatory for u_blitter,
> or since before this series TXF_LZ is optional we continue to support
> it, and set the
> values to 0?
>
> This does the latter, I think it should be fine everywhere.
>
> Dave.


More information about the mesa-dev mailing list