nir/load_store_vectorize: don't ignore subgroup memory barriers
Not sure why I thought this was correct, but we should consider them for
optimization purposes.
Fixes: ce9205c03bd ('nir: add a load/store vectorization pass')
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4202>
(cherry picked from commit f4eb833a12523142d908bf67e08904f4e8a866e9)
diff --git a/.pick_status.json b/.pick_status.json
index 4332bee..62df7c8 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -1957,7 +1957,7 @@
"description": "nir/load_store_vectorize: don't ignore subgroup memory barriers",
"nominated": true,
"nomination_type": 1,
- "resolution": 0,
+ "resolution": 1,
"master_sha": null,
"because_sha": "ce9205c03bd20d26af23ca891e97a9f848a612d1"
},
diff --git a/src/compiler/nir/nir_opt_load_store_vectorize.c b/src/compiler/nir/nir_opt_load_store_vectorize.c
index 6829694..8449be5 100644
--- a/src/compiler/nir/nir_opt_load_store_vectorize.c
+++ b/src/compiler/nir/nir_opt_load_store_vectorize.c
@@ -1196,7 +1196,6 @@
release = nir_intrinsic_memory_semantics(intrin) & NIR_MEMORY_RELEASE;
switch (nir_intrinsic_memory_scope(intrin)) {
case NIR_SCOPE_INVOCATION:
- case NIR_SCOPE_SUBGROUP:
/* a barier should never be required for correctness with these scopes */
modes = 0;
break;