[Mesa-dev] [PATCH] radeonsi: skip DCC render feedback checking if color writes are disabled

[Dieter Nützel]

Tested-by: Dieter Nützel <Dieter at nuetzel-hh.de>

on RX580
with glmark2, UH, UV, Blender, FreeCAD, Krita 4.0.0, KDE Plasma5

Dieter

Am 02.04.2018 05:54, schrieb Marek Olšák:
> From: Marek Olšák <marek.olsak at amd.com>
> 
> ---
>  src/gallium/drivers/radeonsi/si_blit.c          |  5 +++++
>  src/gallium/drivers/radeonsi/si_pipe.h          | 17 +++++++++++++++++
>  src/gallium/drivers/radeonsi/si_state_shaders.c |  6 +-----
>  3 files changed, 23 insertions(+), 5 deletions(-)
> 
> diff --git a/src/gallium/drivers/radeonsi/si_blit.c
> b/src/gallium/drivers/radeonsi/si_blit.c
> index f1c4f6d..c34d677 100644
> --- a/src/gallium/drivers/radeonsi/si_blit.c
> +++ b/src/gallium/drivers/radeonsi/si_blit.c
> @@ -706,20 +706,25 @@ static void
> si_check_render_feedback_resident_images(struct si_context *sctx)
>  		si_check_render_feedback_texture(sctx, tex,
>  						 view->u.tex.level,
>  						 view->u.tex.level,
>  						 view->u.tex.first_layer,
>  						 view->u.tex.last_layer);
>  	}
>  }
> 
>  static void si_check_render_feedback(struct si_context *sctx)
>  {
> +	/* There is no render feedback if color writes are disabled.
> +	 * (e.g. a pixel shader with image stores)
> +	 */
> +	if (!si_get_total_colormask(sctx))
> +		return;
> 
>  	if (!sctx->need_check_render_feedback)
>  		return;
> 
>  	for (int i = 0; i < SI_NUM_SHADERS; ++i) {
>  		si_check_render_feedback_images(sctx, &sctx->images[i]);
>  		si_check_render_feedback_textures(sctx, &sctx->samplers[i]);
>  	}
> 
>  	si_check_render_feedback_resident_images(sctx);
> diff --git a/src/gallium/drivers/radeonsi/si_pipe.h
> b/src/gallium/drivers/radeonsi/si_pipe.h
> index 2053dcb..e75e6c1 100644
> --- a/src/gallium/drivers/radeonsi/si_pipe.h
> +++ b/src/gallium/drivers/radeonsi/si_pipe.h
> @@ -923,11 +923,28 @@ si_htile_enabled(struct r600_texture *tex, 
> unsigned level)
>  	return tex->htile_offset && level == 0;
>  }
> 
>  static inline bool
>  vi_tc_compat_htile_enabled(struct r600_texture *tex, unsigned level)
>  {
>  	assert(!tex->tc_compatible_htile || tex->htile_offset);
>  	return tex->tc_compatible_htile && level == 0;
>  }
> 
> +static inline unsigned si_get_total_colormask(struct si_context *sctx)
> +{
> +	if (sctx->queued.named.rasterizer->rasterizer_discard)
> +		return 0;
> +
> +	struct si_shader_selector *ps = sctx->ps_shader.cso;
> +	unsigned colormask = sctx->framebuffer.colorbuf_enabled_4bit &
> +			     sctx->queued.named.blend->cb_target_mask;
> +
> +	if (!ps->info.properties[TGSI_PROPERTY_FS_COLOR0_WRITES_ALL_CBUFS])
> +		colormask &= ps->colors_written_4bit;
> +	else if (!ps->colors_written_4bit)
> +		colormask = 0; /* color0 writes all cbufs, but it's not written */
> +
> +	return colormask;
> +}
> +
>  #endif
> diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c
> b/src/gallium/drivers/radeonsi/si_state_shaders.c
> index 8fe4c04..45eded9 100644
> --- a/src/gallium/drivers/radeonsi/si_state_shaders.c
> +++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
> @@ -1208,25 +1208,21 @@ static void
> si_shader_selector_key_hw_vs(struct si_context *sctx,
>  	bool ps_disabled = true;
>  	if (ps) {
>  		const struct si_state_blend *blend = sctx->queued.named.blend;
>  		bool alpha_to_coverage = blend && blend->alpha_to_coverage;
>  		bool ps_modifies_zs = ps->info.uses_kill ||
>  				      ps->info.writes_z ||
>  				      ps->info.writes_stencil ||
>  				      ps->info.writes_samplemask ||
>  				      alpha_to_coverage ||
>  				      si_get_alpha_test_func(sctx) != PIPE_FUNC_ALWAYS;
> -
> -		unsigned ps_colormask = sctx->framebuffer.colorbuf_enabled_4bit &
> -					sctx->queued.named.blend->cb_target_mask;
> -		if (!ps->info.properties[TGSI_PROPERTY_FS_COLOR0_WRITES_ALL_CBUFS])
> -			ps_colormask &= ps->colors_written_4bit;
> +		unsigned ps_colormask = si_get_total_colormask(sctx);
> 
>  		ps_disabled = sctx->queued.named.rasterizer->rasterizer_discard ||
>  			      (!ps_colormask &&
>  			       !ps_modifies_zs &&
>  			       !ps->info.writes_memory);
>  	}
> 
>  	/* Find out which VS outputs aren't used by the PS. */
>  	uint64_t outputs_written = vs->outputs_written;
>  	uint64_t inputs_read = 0;