Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-08 Thread Samuel Pitoiset 

LGTM.

Thanks for double checking Tim!

On 5/8/19 1:27 AM, Timothy Arceri wrote:

On 8/5/19 1:51 am, Samuel Pitoiset wrote:

What games are affected btw?


 PERCENTAGE DELTAS    Shaders SGPRs VGPRs SpillSGPR CodeSize 
MaxWaves

 batman-arkham-city   2581 . . . . .
 dawn-of-war-3 244 . . . . .
 f1-2017  5627 . . . . .
 fallout4-vr   196 . . . . .
 nier 1905 . . . . .
 no-mans-sky  4054 . . . . .
 prey 2182 . . . . .
 rot-tomb-raider  8391 . . . . .
 skyrim-vr 494 . . . . .
 sot-tomb-raider   613 . . .   0.01 % .
 the_witcher_3-medium  803 . . . . .
 the_wither_3-ultra   1040    0.05 %   -0.03 % .    . 0.02 %
 valve-vr-pref-trace  323 . . . . .
 wolfenstein-2   1056 . .   -0.20 %  -0.02 % .

-- 


 All affected  69    0.50 %   -0.22 %   -7.69 %  -0.28 %0.57 %
-- 


 Total  29509 . .   -0.03 % . .



Can you please double check before pushing because of the flrp 
changes that landed around?


No change.



On 5/7/19 7:14 AM, Timothy Arceri wrote:

ping!

On 2/5/19 1:38 pm, Timothy Arceri wrote:

The pattern of calling opt algebraic first seems to have originated
in i965. The order in OpenGL drivers generally doesn't matter
because the GLSL IR optimisations do constant folding before
opt algebraic.

However in Vulkan drivers calling opt algebraic first can result
in missed constant folding opportunities.

vkpipeline-db results (VEGA64):

Totals from affected shaders:
SGPRS: 3160 -> 3176 (0.51 %)
VGPRS: 3588 -> 3580 (-0.22 %)
Spilled SGPRs: 52 -> 44 (-15.38 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 12 -> 12 (0.00 %) dwords per thread
Code Size: 261812 -> 261036 (-0.30 %) bytes
LDS: 7 -> 7 (0.00 %) blocks
Max Waves: 346 -> 348 (0.58 %)
Wait states: 0 -> 0 (0.00 %)
---
  src/amd/vulkan/radv_shader.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_shader.c 
b/src/amd/vulkan/radv_shader.c

index cd5a9f2afb4..ad7b2439735 100644
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, 
bool optimize_conservatively,

  NIR_PASS(progress, shader, nir_opt_dead_cf);
  NIR_PASS(progress, shader, nir_opt_cse);
  NIR_PASS(progress, shader, 
nir_opt_peephole_select, 8, true, true);

-    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, 
nir_opt_constant_folding);

+    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_undef);
  NIR_PASS(progress, shader, 
nir_opt_conditional_discard);

  if (shader->options->max_unroll_iterations) {


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-07 Thread Timothy Arceri 

On 8/5/19 1:51 am, Samuel Pitoiset wrote:

What games are affected btw?


 PERCENTAGE DELTASShaders SGPRs VGPRs SpillSGPR CodeSize 
MaxWaves
 batman-arkham-city   2581 . . .. . 

 dawn-of-war-3 244 . . .. . 

 f1-2017  5627 . . .. . 

 fallout4-vr   196 . . .. . 

 nier 1905 . . .. . 

 no-mans-sky  4054 . . .. . 

 prey 2182 . . .. . 

 rot-tomb-raider  8391 . . .. . 

 skyrim-vr 494 . . .. . 

 sot-tomb-raider   613 . . .   0.01 % . 

 the_witcher_3-medium  803 . . .. . 


 the_wither_3-ultra   10400.05 %   -0.03 % .. 0.02 %
 valve-vr-pref-trace  323 . . .. .
 wolfenstein-2   1056 . .   -0.20 %  -0.02 % .

--
 All affected  690.50 %   -0.22 %   -7.69 %  -0.28   %0.57 %
--
 Total  29509 . .   -0.03 % . .



Can you please double check before pushing because of the flrp changes 
that landed around?


No change.



On 5/7/19 7:14 AM, Timothy Arceri wrote:

ping!

On 2/5/19 1:38 pm, Timothy Arceri wrote:

The pattern of calling opt algebraic first seems to have originated
in i965. The order in OpenGL drivers generally doesn't matter
because the GLSL IR optimisations do constant folding before
opt algebraic.

However in Vulkan drivers calling opt algebraic first can result
in missed constant folding opportunities.

vkpipeline-db results (VEGA64):

Totals from affected shaders:
SGPRS: 3160 -> 3176 (0.51 %)
VGPRS: 3588 -> 3580 (-0.22 %)
Spilled SGPRs: 52 -> 44 (-15.38 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 12 -> 12 (0.00 %) dwords per thread
Code Size: 261812 -> 261036 (-0.30 %) bytes
LDS: 7 -> 7 (0.00 %) blocks
Max Waves: 346 -> 348 (0.58 %)
Wait states: 0 -> 0 (0.00 %)
---
  src/amd/vulkan/radv_shader.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
index cd5a9f2afb4..ad7b2439735 100644
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, bool 
optimize_conservatively,

  NIR_PASS(progress, shader, nir_opt_dead_cf);
  NIR_PASS(progress, shader, nir_opt_cse);
  NIR_PASS(progress, shader, nir_opt_peephole_select, 
8, true, true);

-    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_constant_folding);
+    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_undef);
  NIR_PASS(progress, shader, 
nir_opt_conditional_discard);

  if (shader->options->max_unroll_iterations) {


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-07 Thread Samuel Pitoiset 

What games are affected btw?

Can you please double check before pushing because of the flrp changes 
that landed around?


On 5/7/19 7:14 AM, Timothy Arceri wrote:

ping!

On 2/5/19 1:38 pm, Timothy Arceri wrote:

The pattern of calling opt algebraic first seems to have originated
in i965. The order in OpenGL drivers generally doesn't matter
because the GLSL IR optimisations do constant folding before
opt algebraic.

However in Vulkan drivers calling opt algebraic first can result
in missed constant folding opportunities.

vkpipeline-db results (VEGA64):

Totals from affected shaders:
SGPRS: 3160 -> 3176 (0.51 %)
VGPRS: 3588 -> 3580 (-0.22 %)
Spilled SGPRs: 52 -> 44 (-15.38 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 12 -> 12 (0.00 %) dwords per thread
Code Size: 261812 -> 261036 (-0.30 %) bytes
LDS: 7 -> 7 (0.00 %) blocks
Max Waves: 346 -> 348 (0.58 %)
Wait states: 0 -> 0 (0.00 %)
---
  src/amd/vulkan/radv_shader.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
index cd5a9f2afb4..ad7b2439735 100644
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, bool 
optimize_conservatively,

  NIR_PASS(progress, shader, nir_opt_dead_cf);
  NIR_PASS(progress, shader, nir_opt_cse);
  NIR_PASS(progress, shader, nir_opt_peephole_select, 
8, true, true);

-    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_constant_folding);
+    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_undef);
  NIR_PASS(progress, shader, 
nir_opt_conditional_discard);

  if (shader->options->max_unroll_iterations) {


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-07 Thread Bas Nieuwenhuizen 
Nope, r-b

On Tue, May 7, 2019 at 8:36 AM Samuel Pitoiset
 wrote:
>
> Seems fine to,
>
> Reviewed-by: Samuel Pitoiset 
>
> Bas, any comments?
>
> On 5/7/19 7:14 AM, Timothy Arceri wrote:
> > ping!
> >
> > On 2/5/19 1:38 pm, Timothy Arceri wrote:
> >> The pattern of calling opt algebraic first seems to have originated
> >> in i965. The order in OpenGL drivers generally doesn't matter
> >> because the GLSL IR optimisations do constant folding before
> >> opt algebraic.
> >>
> >> However in Vulkan drivers calling opt algebraic first can result
> >> in missed constant folding opportunities.
> >>
> >> vkpipeline-db results (VEGA64):
> >>
> >> Totals from affected shaders:
> >> SGPRS: 3160 -> 3176 (0.51 %)
> >> VGPRS: 3588 -> 3580 (-0.22 %)
> >> Spilled SGPRs: 52 -> 44 (-15.38 %)
> >> Spilled VGPRs: 0 -> 0 (0.00 %)
> >> Private memory VGPRs: 0 -> 0 (0.00 %)
> >> Scratch size: 12 -> 12 (0.00 %) dwords per thread
> >> Code Size: 261812 -> 261036 (-0.30 %) bytes
> >> LDS: 7 -> 7 (0.00 %) blocks
> >> Max Waves: 346 -> 348 (0.58 %)
> >> Wait states: 0 -> 0 (0.00 %)
> >> ---
> >>   src/amd/vulkan/radv_shader.c | 2 +-
> >>   1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
> >> index cd5a9f2afb4..ad7b2439735 100644
> >> --- a/src/amd/vulkan/radv_shader.c
> >> +++ b/src/amd/vulkan/radv_shader.c
> >> @@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, bool
> >> optimize_conservatively,
> >>   NIR_PASS(progress, shader, nir_opt_dead_cf);
> >>   NIR_PASS(progress, shader, nir_opt_cse);
> >>   NIR_PASS(progress, shader, nir_opt_peephole_select,
> >> 8, true, true);
> >> -NIR_PASS(progress, shader, nir_opt_algebraic);
> >>   NIR_PASS(progress, shader, nir_opt_constant_folding);
> >> +NIR_PASS(progress, shader, nir_opt_algebraic);
> >>   NIR_PASS(progress, shader, nir_opt_undef);
> >>   NIR_PASS(progress, shader,
> >> nir_opt_conditional_discard);
> >>   if (shader->options->max_unroll_iterations) {
> >>
> > ___
> > mesa-dev mailing list
> > mesa-dev@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
> ___
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-07 Thread Samuel Pitoiset 

Seems fine to,

Reviewed-by: Samuel Pitoiset 

Bas, any comments?

On 5/7/19 7:14 AM, Timothy Arceri wrote:

ping!

On 2/5/19 1:38 pm, Timothy Arceri wrote:

The pattern of calling opt algebraic first seems to have originated
in i965. The order in OpenGL drivers generally doesn't matter
because the GLSL IR optimisations do constant folding before
opt algebraic.

However in Vulkan drivers calling opt algebraic first can result
in missed constant folding opportunities.

vkpipeline-db results (VEGA64):

Totals from affected shaders:
SGPRS: 3160 -> 3176 (0.51 %)
VGPRS: 3588 -> 3580 (-0.22 %)
Spilled SGPRs: 52 -> 44 (-15.38 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 12 -> 12 (0.00 %) dwords per thread
Code Size: 261812 -> 261036 (-0.30 %) bytes
LDS: 7 -> 7 (0.00 %) blocks
Max Waves: 346 -> 348 (0.58 %)
Wait states: 0 -> 0 (0.00 %)
---
  src/amd/vulkan/radv_shader.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
index cd5a9f2afb4..ad7b2439735 100644
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, bool 
optimize_conservatively,

  NIR_PASS(progress, shader, nir_opt_dead_cf);
  NIR_PASS(progress, shader, nir_opt_cse);
  NIR_PASS(progress, shader, nir_opt_peephole_select, 
8, true, true);

-    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_constant_folding);
+    NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_undef);
  NIR_PASS(progress, shader, 
nir_opt_conditional_discard);

  if (shader->options->max_unroll_iterations) {


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-06 Thread Timothy Arceri 

ping!

On 2/5/19 1:38 pm, Timothy Arceri wrote:

The pattern of calling opt algebraic first seems to have originated
in i965. The order in OpenGL drivers generally doesn't matter
because the GLSL IR optimisations do constant folding before
opt algebraic.

However in Vulkan drivers calling opt algebraic first can result
in missed constant folding opportunities.

vkpipeline-db results (VEGA64):

Totals from affected shaders:
SGPRS: 3160 -> 3176 (0.51 %)
VGPRS: 3588 -> 3580 (-0.22 %)
Spilled SGPRs: 52 -> 44 (-15.38 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 12 -> 12 (0.00 %) dwords per thread
Code Size: 261812 -> 261036 (-0.30 %) bytes
LDS: 7 -> 7 (0.00 %) blocks
Max Waves: 346 -> 348 (0.58 %)
Wait states: 0 -> 0 (0.00 %)
---
  src/amd/vulkan/radv_shader.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
index cd5a9f2afb4..ad7b2439735 100644
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, bool 
optimize_conservatively,
  NIR_PASS(progress, shader, nir_opt_dead_cf);
  NIR_PASS(progress, shader, nir_opt_cse);
  NIR_PASS(progress, shader, nir_opt_peephole_select, 8, true, 
true);
-NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_constant_folding);
+NIR_PASS(progress, shader, nir_opt_algebraic);
  NIR_PASS(progress, shader, nir_opt_undef);
  NIR_PASS(progress, shader, nir_opt_conditional_discard);
  if (shader->options->max_unroll_iterations) {


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH] radv: call constant folding before opt algebraic

2019-05-01 Thread Timothy Arceri 
The pattern of calling opt algebraic first seems to have originated
in i965. The order in OpenGL drivers generally doesn't matter
because the GLSL IR optimisations do constant folding before
opt algebraic.

However in Vulkan drivers calling opt algebraic first can result
in missed constant folding opportunities.

vkpipeline-db results (VEGA64):

Totals from affected shaders:
SGPRS: 3160 -> 3176 (0.51 %)
VGPRS: 3588 -> 3580 (-0.22 %)
Spilled SGPRs: 52 -> 44 (-15.38 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 12 -> 12 (0.00 %) dwords per thread
Code Size: 261812 -> 261036 (-0.30 %) bytes
LDS: 7 -> 7 (0.00 %) blocks
Max Waves: 346 -> 348 (0.58 %)
Wait states: 0 -> 0 (0.00 %)
---
 src/amd/vulkan/radv_shader.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_shader.c b/src/amd/vulkan/radv_shader.c
index cd5a9f2afb4..ad7b2439735 100644
--- a/src/amd/vulkan/radv_shader.c
+++ b/src/amd/vulkan/radv_shader.c
@@ -162,8 +162,8 @@ radv_optimize_nir(struct nir_shader *shader, bool 
optimize_conservatively,
 NIR_PASS(progress, shader, nir_opt_dead_cf);
 NIR_PASS(progress, shader, nir_opt_cse);
 NIR_PASS(progress, shader, nir_opt_peephole_select, 8, true, 
true);
-NIR_PASS(progress, shader, nir_opt_algebraic);
 NIR_PASS(progress, shader, nir_opt_constant_folding);
+NIR_PASS(progress, shader, nir_opt_algebraic);
 NIR_PASS(progress, shader, nir_opt_undef);
 NIR_PASS(progress, shader, nir_opt_conditional_discard);
 if (shader->options->max_unroll_iterations) {
-- 
2.20.1

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev