[Mesa-dev] RFC: mesa/st dynamic sampler support in tgsi

[Marek Olšák]

Using TEMPs instead of address registers seems to be the best choice
here. It will need a CAP though, because some drivers will never
support it. It might be quite a lot of work though.

Marek

On Tue, Aug 5, 2014 at 10:10 PM, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
> *needed*? No, I can't. However without changing a bunch of stuff
> around, doing something like
>
> TEX TEMP[ADDR[0].x], SAMP[reladdr], coord, etc
>
> Is going to be tricky. It'd be a little dangerous to use ADDR[1] there
> since it's meant to be the dimension, and you could easily have, with
> tessellation,
>
> TEX OUT[ADDR[1].x][ADDR[0].x], SAMP[reladdr], ...
>
> Now I could obviously rewrite that as
>
> TEX TEMP[foo], SAMP[ADDR[0].x], ...
> UARL ADDR[0]
> UARL ADDR[1]
> MOV OUT[ADDR[1].x][ADDR[0].x], TEMP[foo]
>
> But then I'd have to somehow communicate the emit() that it needs to
> arrange for this. Not impossible, but tricky.
>
>   -ilia
>
> On Tue, Aug 5, 2014 at 4:04 PM, Marek Olšák <maraeo at gmail.com> wrote:
>> Could you please show us an example in TGSI where a 3rd address
>> register is needed?
>>
>> Thanks,
>>
>> Marek
>>
>> On Tue, Aug 5, 2014 at 6:54 AM, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
>>> Hello,
>>>
>>> I'm in the process of working out the details of adding dynamic
>>> sampler support to mesa/st as part of ARB_gpu_shader5. Unfortunately
>>> I've run into a bit of a roadblock.
>>>
>>> One approach I've tried is to load the indirect sampler index into an
>>> address register and then use that as the offset. However in order to
>>> realistically support that I need a 3rd address register, while right
>>> now tgsi_ureg limits it to 2 address registers. Would it be OK to
>>> relax the restriction to 3? (The reason I need a third is that if e.g.
>>> the dst has a reladdr, I'd have to change a bunch of stuff to make
>>> that function properly without a 3rd addr reg, since I can't just MOV
>>> the sampler to a temp like it does for all the other src's.)
>>>
>>> Another apporach I've tried is to just use a TEMP register as the
>>> indirect offset here. Unfortunately this gets destroyed by
>>> st_glsl_to_tgsi's various optimizations which assume that temp
>>> registers can't be reladdr's and so messes up the values. I started
>>> adding support for that, but then quickly realized that was probably
>>> not the right thing to do.
>>>
>>> So... should I increase the number of address registers to 1? Or is
>>> there some other simple approach that I'm missing?
>>>
>>> Thanks,
>>>
>>>   -ilia
>>> _______________________________________________
>>> mesa-dev mailing list
>>> mesa-dev at lists.freedesktop.org
>>> http://lists.freedesktop.org/mailman/listinfo/mesa-dev