zhiics edited a comment on pull request #6068: URL: https://github.com/apache/incubator-tvm/pull/6068#issuecomment-659071984
A better implementation should be invoking a `target` specific optimization after partitioning. This is more to make #5915 clean. The reason is what @comaniac mentioned above. We should think about all these in the RFC. The flow would be the following: target required processing passes -> annotation -> merge regions -> partitioning -> target required post processing passes -> relay compilation/codegen and target codegen ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected]
