fix(gpu): retain resources across entry pragmas

Fixes #6651
This commit is contained in:
morluto
2026-10-23 19:17:42 +08:00
parent 8695ca8257
commit 42086fdba7
4 changed files with 92 additions and 1 deletions
+3 -1
View File
@@ -82,7 +82,9 @@ Kernel representations have different meanings:
supported static declarations inside that entry; they do not establish module
shared memory, dynamic allocation or physical device resources.
PTX 1.0–1.3 declares entry parameters in the body; PTX 1.4 and later uses the
entry parameter list. A `.target compute_xx` spelling is retained as written;
entry parameter list. Entry-scoped `.pragma` statements before the body do
not end the declaration or suppress its body resources. A `.target
compute_xx` spelling is retained as written;
NVIDIA PTX defines it as a synonym for `.target sm_xx` when comparing its
target identity with the fatbin declaration. `.callprototype` formal
parameters are excluded from entry virtual-register counts. Parameter names