Back out "Flip triton kernel default layout constraint to "needs_fixed_stride_order" (#135581)"

tissue3 · facebook-github-bot · commit 135a22be06a4 · 2024-09-16T12:20:17.000-07:00
Test Plan: make train-hstu-cint-publish-bf16-tgif-local

Differential Revision: D62766335
diff --git a/torch/_inductor/config.py b/torch/_inductor/config.py
@@ -74,7 +74,7 @@ def autotune_remote_cache_default() -> Optional[bool]:
 
 # The default layout constraint for user-defined triton kernels.
 # See "The default layout constraint for custom operators" for options.
-triton_kernel_default_layout_constraint = "needs_fixed_stride_order"
+triton_kernel_default_layout_constraint = "flexible_layout"
 
 # use cpp wrapper instead of python wrapper
 cpp_wrapper = os.environ.get("TORCHINDUCTOR_CPP_WRAPPER", "0") == "1"
