mbs-octoml commented on pull request #8835: URL: https://github.com/apache/tvm/pull/8835#issuecomment-905558007
Are the hash and equality on Targets legit? Also, possibly relevant, there's a bit of hackery to force shape functions to end up on the 'cpu' target which is hard coded, so it could be there's two 'llvm' targets in action that look the same but are distinct objects as far has hash/equality is concerned. I'd like to fix that by making device planning responsible for spelling out everything so there's no special cases left here. Note ideally the code would be: lowered_functions[target]->Add(...) ie let the defaut ctor do it's thing if the key is not present. -- 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: commits-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org