tqchen commented on PR #18877: URL: https://github.com/apache/tvm/pull/18877#issuecomment-4011338420
Thanks @mitiskuma i think only thing left is to guard on maximum staging buffer being used and be able to flush to avoid unbounded use-cases. Indeed that most LLM decoding use-cases won't have such issue, but we want runtime implementation to generally be robust to all possible usecase while optimized for usecases we have in mind -- 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]
