tqchen commented on PR #18871: URL: https://github.com/apache/tvm/pull/18871#issuecomment-4000184002
thanks @mitiskuma , mainly from design pov: - A0: i think we should be able to get lazy + staging buffer in, these are great changes and i think can get polished version and non-controversial. - A1: The caching logic i think can needs a bit deliberation, main in terms of how we ideally structure it. If we have something like an explicit CacheState and clearly think through some of the hash collison scenario to make them robust, i think we can also bring some in. From the operational pov, likely splitting the two into two PRs would make sense. Likely we can focus on landing A0 soon, then working on A1 if we find it being useful -- 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]
