radv: fix computing the pervertex LDS size with NGG streamout

The NGG streamout lowering pass allocates space for all outputs which
means we have to align our computation. Otherwise, the maximum number
of vertices is incorrect and we end up by reaching the maximum allowed
LDS size. This code could be shared instead of being duplicated but
that's for later.

Fixes some transform feedback tests with Zink and
RADV_PERFTEST=ngg_streamout on GFX10.3.

Cc: 22.3 mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19801>

diff --git a/src/amd/vulkan/radv_shader_info.c b/src/amd/vulkan/radv_shader_info.c
index 8e8ec56..3625fe2 100644 (file)
--- a/src/amd/vulkan/radv_shader_info.c
+++ b/src/amd/vulkan/radv_shader_info.c
@@ -1048,8 +1048,13 @@ gfx10_get_ngg_info(const struct radv_device *device, struct radv_pipeline_stage
       /* LDS size for passing data from GS to ES. */
       struct radv_streamout_info *so_info = &es_info->so;
 
-      if (so_info->num_outputs)
-         esvert_lds_size = 4 * so_info->num_outputs + 1;
+      if (so_info->num_outputs) {
+         /* Compute the same pervertex LDS size as the NGG streamout lowering pass which allocates
+          * space for all outputs.
+          * TODO: only alloc space for outputs that really need streamout.
+          */
+         esvert_lds_size = 4 * es_stage->nir->num_outputs + 1;
+      }
 
       /* GS stores Primitive IDs (one DWORD) into LDS at the address
        * corresponding to the ES thread of the provoking vertex. All