tqchen commented on issue #5307: [TIR] Make lower_warp_memory support extent(threadIdx.x) < warp_size URL: https://github.com/apache/incubator-tvm/pull/5307#issuecomment-612561970 I actually meant the "bind to warp scope mode". What worries me a bit is that in our original convention, threadIdx.x always means warp index, and the previous comment was about whether or not we should keep threadIdx.x still equal to be warp size, but to detect patttern ```A[wi] = B[(wi/4)*4+ ((wi % 4) +1) %4] ``` (where wi = threadIdx.x) for the half shuffle case
---------------------------------------------------------------- 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] With regards, Apache Git Services
