On Saturday, April 16, 2016 4:43:00 PM PDT Topi Pohjolainen wrote:
> I noticed using one synthetic benchmark a sequence where hiz depth
> operations are run first followed by a color buffer clear. In such
> case there is a difference between blorp and meta clear where meta
> configures L3 but blorp doesn't. I didn't see any problems in
> practise without the configure but this is the right thing to do.
>
> Signed-off-by: Topi Pohjolainen <topi.pohjolai...@intel.com>
> ---
> src/mesa/drivers/dri/i965/gen7_blorp.cpp | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/src/mesa/drivers/dri/i965/gen7_blorp.cpp b/src/mesa/drivers/
dri/i965/gen7_blorp.cpp
> index ef96ec5..b85a6a5 100644
> --- a/src/mesa/drivers/dri/i965/gen7_blorp.cpp
> +++ b/src/mesa/drivers/dri/i965/gen7_blorp.cpp
> @@ -802,6 +802,7 @@ gen7_blorp_exec(struct brw_context *brw,
> params->dst.num_samples > 1 ?
> (1 << params->dst.num_samples) - 1 : 1);
> gen6_blorp_emit_state_base_address(brw, params);
> + gen7_l3_state.emit(brw);
> gen6_blorp_emit_vertices(brw, params);
> gen7_blorp_emit_urb_config(brw);
> if (params->use_wm_prog) {
> Hmm...I'm not sure about this. It configures the L3 based on the GL context's active shader stages...which are not the shaders that BLORP will be using. We could instead call gen7_restore_default_l3_config(). However...I suspect that pretty much any L3 configuration the driver sets up will work for BLORP's purposes. If we're just worried about BLORP being the first operation (before normal drawing), it might make sense to have brw_upload_initial_gpu_state() call gen7_restore_default_l3_config(). Then just let BLORP inherit the last state we programmed. I don't know exactly what the kernel guarantees, though. It's probably wise to check with Curro on what he thinks the right approach is.
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
