[Mesa-dev] [PATCH 2/3] radv: hardcode shader WAVE_LIMIT to the maximum value

[Samuel Pitoiset]

On 8/14/18 5:46 PM, Marek Olšák wrote:
> 
> 
> On Tue, Aug 14, 2018, 8:28 AM Samuel Pitoiset <samuel.pitoiset at gmail.com 
> <mailto:samuel.pitoiset at gmail.com>> wrote:
> 
> 
> 
>     On 8/1/18 5:12 AM, Marek Olšák wrote:
>      > On Fri, Oct 20, 2017 at 4:34 PM, Andres Rodriguez
>     <andresx7 at gmail.com <mailto:andresx7 at gmail.com>> wrote:
>      >> When WAVE_LIMIT is set, a submission will opt-in for SPI based
>     resource
>      >> scheduling. Because this mechanism is cooperative, we must
>     ensure that
>      >> all submissions have this field set, otherwise they will bypass
>     resource
>      >> arbitration.
>      >>
>      >> We always hardcode the field to its maximum value, instead of
>      >> attempting to calculate an approximate usage. In testing, there
>     were no
>      >> benefits to using anything other than the maximum.
>      >>
>      >> Signed-off-by: Andres Rodriguez <andresx7 at gmail.com
>     <mailto:andresx7 at gmail.com>>
>      >> ---
>      >>   src/amd/vulkan/si_cmd_buffer.c          | 27
>     ++++++++++++++++++---------
>      >>   src/gallium/drivers/radeonsi/si_state.c | 21 ++++++++++++++-------
>      >>   2 files changed, 32 insertions(+), 16 deletions(-)
>      >>
>      >> diff --git a/src/amd/vulkan/si_cmd_buffer.c
>     b/src/amd/vulkan/si_cmd_buffer.c
>      >> index de3b388..ac3dff8 100644
>      >> --- a/src/amd/vulkan/si_cmd_buffer.c
>      >> +++ b/src/amd/vulkan/si_cmd_buffer.c
>      >> @@ -179,7 +179,8 @@ si_emit_compute(struct radv_physical_device
>     *physical_device,
>      >>          radeon_emit(cs, 0);
>      >>          radeon_emit(cs, 0);
>      >>
>      >> -       radeon_set_sh_reg_seq(cs,
>     R_00B854_COMPUTE_RESOURCE_LIMITS, 3);
>      >> +       radeon_set_sh_reg_seq(cs, R_00B854_COMPUTE_RESOURCE_LIMITS,
>      >> +                             S_00B854_WAVES_PER_SH(0x3));
>      >
>      > This part doesn't set WAVES_PER_SH. Also, the number 3 would be
>     wrong for it.
> 
>     This is just a default value for that register. COMPUTE_RESOURCE_LIMITS
>     is computed like RadeonSI when we generate the PM4 stuff for compute
>     pipelines, and then emitted at dispatch time.
> 
>     Do you think we should remove that default value?
> 
> 
> Please read the patch again. It's obvious if you know what the seq 
> function does.

Oh right, this is totally wrong...

> 
> Marek
> 
> 
>      >
>      > Marek
>      > _______________________________________________
>      > mesa-dev mailing list
>      > mesa-dev at lists.freedesktop.org
>     <mailto:mesa-dev at lists.freedesktop.org>
>      > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
>      >
>