[Mesa-dev] [PATCH 3/3] i965/fs: Drop no-op shifts involving 0.

[Matt Turner]

On Fri, Oct 25, 2013 at 2:49 PM, Eric Anholt <eric at anholt.net> wrote:
> I noticed this in a shader in Unigine Heaven that was spilling.  While it
> doesn't really reduce register pressure, it shaves a few instructions
> anyway (7955 -> 7882).
>
> v2: Fix turning "0 >> x" into "x" instead of "0" (caught by Erik
>     Faye-Lund).
> ---
>  src/glsl/opt_algebraic.cpp | 10 ++++++++++
>  1 file changed, 10 insertions(+)
>
> diff --git a/src/glsl/opt_algebraic.cpp b/src/glsl/opt_algebraic.cpp
> index 2e33dfe..a07e153 100644
> --- a/src/glsl/opt_algebraic.cpp
> +++ b/src/glsl/opt_algebraic.cpp
> @@ -346,6 +346,16 @@ ir_algebraic_visitor::handle_expression(ir_expression *ir)
>        }
>        break;
>
> +   case ir_binop_rshift:
> +   case ir_binop_lshift:
> +      /* 0 >> x == 0 */
> +      if (is_vec_zero(op_const[0]))
> +         return ir->operands[0];

Any value to writing ir_constant::zero(ir, ir->type) here instead?

Either way, this series is

Reviewed-by: Matt Turner <mattst88 at gmail.com>

... for whatever that's worth these days. :)

I do think these clean ups make the code a lot clearer. Nice.