Re: [I] [Feature] Support specifying CUDA version [tvm-ffi]
junrushao commented on issue #430: URL: https://github.com/apache/tvm-ffi/issues/430#issuecomment-3903356369 A related issue: https://github.com/apache/tvm-ffi/pull/440. Could you all suggest actionable plan? -- 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]
Re: [I] [Feature] Support specifying CUDA version [tvm-ffi]
tqchen commented on issue #430: URL: https://github.com/apache/tvm-ffi/issues/430#issuecomment-3872213062 sounds good, maybe we can suggest `TVM_FFI_CUDA_ARCH_LIST` first then update the api if needed, whatever the path we think convenient for the user side. -- 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]
Re: [I] [Feature] Support specifying CUDA version [tvm-ffi]
DarkSharpness commented on issue #430: URL: https://github.com/apache/tvm-ffi/issues/430#issuecomment-3872031876 I just realized `TVM_FFI_CUDA_ARCH_LIST` can fit this need by 1. overiding the env var 2. build JIT kernel 3. reset the env var. This brings a little inconvenience for user side, but can fit in every case. I'm not sure whether we really need a special argument to override the version. -- 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]
Re: [I] [Feature] Support specifying CUDA version [tvm-ffi]
tqchen commented on issue #430: URL: https://github.com/apache/tvm-ffi/issues/430#issuecomment-3871533222 @DarkSharpness would `TVM_FFI_CUDA_ARCH_LIST` fit this need? or are you suggesting we pass them as arguments in the API. Both should be fine approaches, and i think we would be happy to take a pr -- 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]
