[Mesa-dev] [PATCH 1/2] i965: Use nir_lower_load_const_to_scalar().

Eduardo Lima Mitev elima at igalia.com
Fri Jan 22 08:22:27 PST 2016


On 01/22/2016 01:37 AM, Kenneth Graunke wrote:
> I don't know why, but we never hooked up this pass Eric wrote.
> Otherwise, you can end up with stupid scalarized code such as:
> 
>    vec4 ssa_7 = load_const (0.0, 0.0, 0.0, 0.0)
>    vec4 ssa_8 = ...
>    vec1 ssa_9 = feq ssa_8, ssa_7
>    vec1 ssa_10 = feq ssa_8.y, ssa_7.y
>    vec1 ssa_11 = feq ssa_8, ssa_7.z
>    vec1 ssa_12 = feq ssa_8.y, ssa_7.w
> 
> ssa_8.xyxy == <0, 0, 0, 0> should only take two feq instructions.
> 
> shader-db on Skylake:
> 
> total instructions in shared programs: 9111788 -> 9111384 (-0.00%)
> instructions in affected programs: 32421 -> 32017 (-1.25%)
> helped: 277
> HURT: 69
> 
> total cycles in shared programs: 69221226 -> 69219394 (-0.00%)
> cycles in affected programs: 917796 -> 915964 (-0.20%)
> helped: 317
> HURT: 408
> 

I get similar results in HSW:

total instructions in shared programs: 6256788 -> 6256504 (-0.00%)
instructions in affected programs: 26424 -> 26140 (-1.07%)
helped: 240
HURT: 0

total cycles in shared programs: 56009352 -> 56010506 (0.00%)
cycles in affected programs: 807916 -> 809070 (0.14%)
helped: 245
HURT: 325

I wonder why the pass was not enabled before.

Patch is:

Reviewed-by: Eduardo Lima Mitev <elima at igalia.com>

> This also prevents regressions when disabling channel expressions.
> 
> Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
> ---
>  src/mesa/drivers/dri/i965/brw_nir.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/src/mesa/drivers/dri/i965/brw_nir.c b/src/mesa/drivers/dri/i965/brw_nir.c
> index 935529a..ce9b9db 100644
> --- a/src/mesa/drivers/dri/i965/brw_nir.c
> +++ b/src/mesa/drivers/dri/i965/brw_nir.c
> @@ -482,6 +482,11 @@ brw_preprocess_nir(nir_shader *nir, bool is_scalar)
>  
>     nir = nir_optimize(nir, is_scalar);
>  
> +   if (is_scalar) {
> +      OPT_V(nir_lower_load_const_to_scalar);
> +      OPT(nir_opt_cse);
> +   }
> +
>     /* Lower a bunch of stuff */
>     OPT_V(nir_lower_var_copies);
>  
> 



More information about the mesa-dev mailing list