tqchen commented on PR #496: URL: https://github.com/apache/tvm-ffi/pull/496#issuecomment-3994090231
thanks @Aphoh ! the later two makes sense and likely it means we need a Stream wrapper for users so they don't have to always deal with unsafe with stream. Would be good to look into fixing the tests. The mut Tensor part also makes sense, but likely that could make structuring kernels slightly harder since we need unique mut Tensor for intermediates as most are read writes, we could try to first get it in. As most common pattern in low-level kernels are destination passing. But most underlying kernels exposed by tvm-ffi still do not follow the mut rule (which indeed it more unsafe due to concurrent write) -- 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]
