[Mesa-dev] [PATCH v2] main: memcpy larger chunks in _mesa_propagate_uniforms_to_driver_storage

[Nicolai Hähnle]

Pushed, thanks!

On 22.07.2016 13:10, Nils Wallménius wrote:
> When possible, do the memcpy on larger blocks. This reduces cycles
> spent in _mesa_propagate_uniforms_to_driver_storage from
> 1.51 % to 0.62% according to perf during the Unigine Heaven benchmark.
> It did not affect the framerate of the benchmark. The system used for
> testing was an i5 6600K with a Radeon R9 380.
>
> Piglit hangs randomly on this system both with and without the patch
> so i could not make a comparison.
>
> v2: fixed whitespace
>
> Signed-off-by: Nils Wallménius <nils.wallmenius at gmail.com>
> Reviewed-by: Nicolai Hähnle <nicolai.haehnle at amd.com>
> ---
>  src/mesa/main/uniform_query.cpp | 29 +++++++++++++++++++++++------
>  1 file changed, 23 insertions(+), 6 deletions(-)
>
> diff --git a/src/mesa/main/uniform_query.cpp b/src/mesa/main/uniform_query.cpp
> index ab22a0e..b9b9ff2 100644
> --- a/src/mesa/main/uniform_query.cpp
> +++ b/src/mesa/main/uniform_query.cpp
> @@ -578,14 +578,31 @@ _mesa_propagate_uniforms_to_driver_storage(struct gl_uniform_storage *uni,
>  	 unsigned j;
>  	 unsigned v;
>
> -	 for (j = 0; j < count; j++) {
> -	    for (v = 0; v < vectors; v++) {
> -	       memcpy(dst, src, src_vector_byte_stride);
> -	       src += src_vector_byte_stride;
> -	       dst += store->vector_stride;
> +	 if (src_vector_byte_stride == store->vector_stride) {
> +	    if (extra_stride) {
> +	       for (j = 0; j < count; j++) {
> +	          memcpy(dst, src, src_vector_byte_stride * vectors);
> +	          src += src_vector_byte_stride * vectors;
> +	          dst += store->vector_stride * vectors;
> +
> +	          dst += extra_stride;
> +	       }
> +	    } else {
> +	       /* Unigine Heaven benchmark gets here */
> +	       memcpy(dst, src, src_vector_byte_stride * vectors * count);
> +	       src += src_vector_byte_stride * vectors * count;
> +	       dst += store->vector_stride * vectors * count;
>  	    }
> +	 } else {
> +	    for (j = 0; j < count; j++) {
> +	       for (v = 0; v < vectors; v++) {
> +	          memcpy(dst, src, src_vector_byte_stride);
> +	          src += src_vector_byte_stride;
> +	          dst += store->vector_stride;
> +	       }
>
> -	    dst += extra_stride;
> +	       dst += extra_stride;
> +	    }
>  	 }
>  	 break;
>        }
>