add llvm compile arg for dtk2310

4c972a6e · aiss · 3fa19ffc · 4c972a6e
Commit 4c972a6e authored Oct 12, 2023 by aiss
Show whitespace changes
Inline Side-by-side

Showing with 2 additions and 1 deletion

op_builder/builder.py op_builder/builder.py +2 -1

No files found.
--- a/op_builder/builder.py
+++ b/op_builder/builder.py
@@ -681,7 +681,8 @@ class CUDAOpBuilder(OpBuilder):
                '-U__HIP_NO_HALF2_OPERATORS__',
                '-DROCM_VERSION_MAJOR=%s' % ROCM_MAJOR,
                '-DROCM_VERSION_MINOR=%s' % ROCM_MINOR,
-                '--gpu-max-threads-per-block=1024'
+                '--gpu-max-threads-per-block=1024',
+                '-mllvm -amdgpu-enable-flat-scratch=false'
            ]
        else:
            cuda_major, _ = installed_cuda_version()