Lunderberg commented on code in PR #15809:
URL: https://github.com/apache/tvm/pull/15809#discussion_r1338735257
##########
python/tvm/relax/vm_build.py:
##########
@@ -310,6 +310,7 @@ def foo(x: Tensor((3, 4), "float32"), y: Tensor((3, 4),
"float32")):
passes.append(relax.transform.RemovePurityChecking())
passes.append(relax.transform.CallTIRRewrite())
passes.append(relax.transform.StaticPlanBlockMemory())
+ passes.append(relax.transform.LowerAllocTensor())
if
tvm.transform.PassContext.current().config.get("relax.backend.use_cuda_graph",
False):
passes.append(relax.transform.RewriteCUDAGraph())
Review Comment:
Thank you on the reminder. I had assumed that the tests in
`tests/python/relax/test_transform_rewrite_cuda_graph.py` and
`tests/python/relax/test_vm_cuda_graph.py` would be sufficient. On closer
inspection, it turns out that the former provides the input of
`RewriteCUDAGraph`, the latter provides the output from `RewriteCUDAGraph`, and
neither test the behavior of the pass as it exists within a lowering flow. On
writing a quick end-to-end test, there is an issue that occurs within the
cudagraph rewriting pass.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
