Module: Mesa
Branch: main
Commit: 2011b678c538b451eef5a1986adee7e25d6ee04f
URL:
http://cgit.freedesktop.org/mesa/mesa/commit/?id=2011b678c538b451eef5a1986adee7e25d6ee04f
Author: Rhys Perry <pendingchao...@gmail.com>
Date: Fri Nov 3 15:25:48 2023 +0000
radv: disable gs_fast_launch=2 by default
This currently causes severe performance problems in Remnant 2 and
Alan Wake 2.
Signed-off-by: Rhys Perry <pendingchao...@gmail.com>
Reviewed-by: Timur Kristóf <timur.kris...@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
Fixes: 28ebe62af2cb ("radv: enable mesh shader gs_fast_launch=2 and multi-row
export")
Gitlab: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10071
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26090>
---
docs/envvars.rst | 2 ++
src/amd/vulkan/radv_debug.h | 1 +
src/amd/vulkan/radv_device.c | 3 ++-
src/amd/vulkan/radv_instance.c | 1 +
src/amd/vulkan/radv_pipeline.c | 5 +++++
src/amd/vulkan/radv_shader.h | 1 +
6 files changed, 12 insertions(+), 1 deletion(-)
diff --git a/docs/envvars.rst b/docs/envvars.rst
index 9e225a3c2be..b46b070a13c 100644
--- a/docs/envvars.rst
+++ b/docs/envvars.rst
@@ -1342,6 +1342,8 @@ RADV driver environment variables
enable wave64 for ray tracing shaders (GFX10+)
``video_decode``
enable experimental video decoding support
+ ``gsfastlaunch2``
+ use GS_FAST_LAUNCH=2 for Mesh shaders (GFX11+)
.. envvar:: RADV_TEX_ANISO
diff --git a/src/amd/vulkan/radv_debug.h b/src/amd/vulkan/radv_debug.h
index 82d93f4e196..d28c2de8043 100644
--- a/src/amd/vulkan/radv_debug.h
+++ b/src/amd/vulkan/radv_debug.h
@@ -88,6 +88,7 @@ enum {
RADV_PERFTEST_RT_WAVE_64 = 1u << 10,
RADV_PERFTEST_VIDEO_DECODE = 1u << 11,
RADV_PERFTEST_DMA_SHADERS = 1u << 12,
+ RADV_PERFTEST_GS_FAST_LAUNCH_2 = 1u << 13,
};
bool radv_init_trace(struct radv_device *device);
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index f3c92ed7197..c6ada5e8354 100644
--- a/src/amd/vulkan/radv_device.c
+++ b/src/amd/vulkan/radv_device.c
@@ -943,7 +943,8 @@ radv_CreateDevice(VkPhysicalDevice physicalDevice, const
VkDeviceCreateInfo *pCr
device->pbb_allowed =
device->physical_device->rad_info.gfx_level >= GFX9 &&
!(device->instance->debug_flags & RADV_DEBUG_NOBINNING);
- device->mesh_fast_launch_2 = device->physical_device->rad_info.gfx_level >=
GFX11;
+ device->mesh_fast_launch_2 = (device->instance->perftest_flags &
RADV_PERFTEST_GS_FAST_LAUNCH_2) &&
+ device->physical_device->rad_info.gfx_level >=
GFX11;
/* The maximum number of scratch waves. Scratch space isn't divided
* evenly between CUs. The number is only a function of the number of CUs.
diff --git a/src/amd/vulkan/radv_instance.c b/src/amd/vulkan/radv_instance.c
index 4757fe89949..abc410e33b8 100644
--- a/src/amd/vulkan/radv_instance.c
+++ b/src/amd/vulkan/radv_instance.c
@@ -99,6 +99,7 @@ static const struct debug_control radv_perftest_options[] =
{{"localbos", RADV_P
{"rtwave64",
RADV_PERFTEST_RT_WAVE_64},
{"video_decode",
RADV_PERFTEST_VIDEO_DECODE},
{"dmashaders",
RADV_PERFTEST_DMA_SHADERS},
+ {"gsfastlaunch2",
RADV_PERFTEST_GS_FAST_LAUNCH_2},
{NULL, 0}};
const char *
diff --git a/src/amd/vulkan/radv_pipeline.c b/src/amd/vulkan/radv_pipeline.c
index 5b203a16235..d6581e13451 100644
--- a/src/amd/vulkan/radv_pipeline.c
+++ b/src/amd/vulkan/radv_pipeline.c
@@ -214,6 +214,11 @@ radv_generate_pipeline_key(const struct radv_device
*device, const VkPipelineSha
key.vertex_robustness1 = 1u;
}
+ for (uint32_t i = 0; i < num_stages; i++) {
+ if (stages[i].stage == VK_SHADER_STAGE_MESH_BIT_EXT &&
device->mesh_fast_launch_2)
+ key.mesh_fast_launch_2 = 1u;
+ }
+
return key;
}
diff --git a/src/amd/vulkan/radv_shader.h b/src/amd/vulkan/radv_shader.h
index f902d3e1e94..0582e6b1e47 100644
--- a/src/amd/vulkan/radv_shader.h
+++ b/src/amd/vulkan/radv_shader.h
@@ -99,6 +99,7 @@ struct radv_pipeline_key {
uint32_t mesh_shader_queries : 1;
uint32_t vertex_robustness1 : 1;
+ uint32_t mesh_fast_launch_2 : 1;
struct radv_shader_stage_key stage_info[MESA_VULKAN_SHADER_STAGES];
