vinx13 commented on issue #13850: URL: https://github.com/apache/tvm/issues/13850#issuecomment-1405946011
Tensorization approach is also feasible. You can write a tensor intrin implementing the above formula and mixing ordinary * and llvm intrinsics. You will need to write a vectorized loop of multiply and check the generated llvm code is expected. Alternatively you can implement micro kernel in asm or llvm byte code see https://github.com/apache/tvm/issues/1276. -- 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]
