tqchen commented on PR #18871: URL: https://github.com/apache/tvm/pull/18871#issuecomment-3999797916
thanks for the PR. Given GPU is async, i think it is relatively ok and desirable to keep runtime simple and not containing too smart of a caching in this case. This being said, lazy submission could make sense, although would be good to see the performance measurement. Since eager also have its own merit as host compute can run in parallel with gpu compute -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
