[Piglit] [PATCH] glean: Skip depth/stencil performance test in quick mode.

[Brian Paul]

On 08/02/2011 10:25 AM, Eric Anholt wrote:
> I don't understand why glean would have performance microbenchmarks in
> it, but I haven't ever seen anyone interested in the results.  Cuts
> many seconds of an absurdly huge window showing up on my desktop.

Back when I was working on the Chromium project we were interested in 
the readback rate for reading the depth (and color) buffers.  I used 
this test to measure the performance of various cards.


> ---
>   tests/glean/tdepthstencil.cpp |    6 +++++-
>   1 files changed, 5 insertions(+), 1 deletions(-)
>
> diff --git a/tests/glean/tdepthstencil.cpp b/tests/glean/tdepthstencil.cpp
> index ad672ef..c8c1ca9 100644
> --- a/tests/glean/tdepthstencil.cpp
> +++ b/tests/glean/tdepthstencil.cpp
> @@ -303,7 +303,7 @@ DepthStencilTest::runOne(DepthStencilResult&r, Window&w)
>   		r.pass = testDrawAndRead();
>   	if (r.pass)
>   		r.pass = testTextureOperations();
> -	if (r.pass)
> +	if (r.pass&&  !env->options.quick)
>   		testPerformance(r);
>   }
>
> @@ -318,6 +318,10 @@ DepthStencilTest::logOne(DepthStencilResult&r)
>   		char str[1000];
>   		double mbps;
>
> +		/* Skip perf results if we didn't test that. */
> +		if (env->options.quick)
> +			return;
> +
>   		env->log<<  "\tglReadPixels GL_DEPTH_STENCIL rate: ";
>   		mbps = r.readDepthStencilRate * sizeof(GLuint) / (1024*1024);
>   		sprintf(str, "%.2f", mbps);

Reviewed-by: Brian Paul <brianp at vmware.com>