-
Notifications
You must be signed in to change notification settings - Fork 3.8k
[TIR] Fix reverse_compute_at for trivial region with trivial block var #11234
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. Can confirm that this PR fixed my repro on meta_schedule with LLVM. Thanks! @vinx13. BTW do you have any insight on why this issue is only happening on LLVM but not CUDA?
|
CUDA will call |
zxybazh
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. Thanks for the quick fix and thanks to @shingjan for reporting this issue!
5fba48d to
42231b8
Compare
|
Reading the cryptic error message, it seems that the |
In |
|
That makes a lot of sense @wrongtest |
apache#11234) * [TIR] Fix reverse_compute_at for trivial region with trivial block var * Prevent handle arithmetics
apache#11234) * [TIR] Fix reverse_compute_at for trivial region with trivial block var * Prevent handle arithmetics
This PR fixed
compute_at/reverse_compute_atwhen block access regions contain constants (e.g.A[0, vi, vj]).Background:
#11187 added additional simplification when creating prim func from TE. This may result in block var with constant binding values simplified.
cc @junrushao1994 @spectrometerHBH @zxybazh @shingjan