Mesa (master): radv: fix performance regression by restoring TC-compat HTILE in GENERAL

[GitLab Mirror]

Module: Mesa
Branch: master
Commit: f40a7d3c93d186b3415227945fd2df00f81c7f4a
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=f40a7d3c93d186b3415227945fd2df00f81c7f4a

Author: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Date:   Wed Jan  6 16:19:21 2021 +0100

radv: fix performance regression by restoring TC-compat HTILE in GENERAL

This fixes a performance regression for games (eg. Youngblood) that
declare all images as concurrent. This is likely buggy for compute
queues but this just restores the previous behaviour for now.

Fixes: f4f096805b0 ("radv: fix TC-compat HTILE images with DST_OPTIMAL on the compute queue")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8351>

---

 src/amd/vulkan/radv_image.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/amd/vulkan/radv_image.c b/src/amd/vulkan/radv_image.c
index ff51a07d10e..c9f247a5029 100644
--- a/src/amd/vulkan/radv_image.c
+++ b/src/amd/vulkan/radv_image.c
@@ -1774,8 +1774,11 @@ bool radv_layout_is_htile_compressed(const struct radv_device *device,
 		 * depth pass because this allows compression and this reduces
 		 * the number of decompressions from/to GENERAL.
 		 */
+		/* FIXME: Enabling TC-compat HTILE in GENERAL on the compute
+		 * queue is likely broken for eg. depth/stencil copies.
+		 */
 		if (radv_image_is_tc_compat_htile(image) &&
-		    queue_mask == (1u << RADV_QUEUE_GENERAL) &&
+		    queue_mask & (1u << RADV_QUEUE_GENERAL) &&
 		    !in_render_loop) {
 			/* GFX10+ supports compressed writes to HTILE. */
 			return device->physical_device->rad_info.chip_class >= GFX10 ||