Kewen & David, Thanks for your comments. 在 2023/10/17 10:19, Kewen.Lin 写道: > I think David raised a good question, it sounds to me that the current > handling simply consider that if MOVE_MAX_PIECES is set to 16, the > required operations for this optimization on TImode are always available, > but unfortunately on rs6000 the assumption doesn't hold, so could we > teach generic code instead?
Finally I found that it doesn't check if the scalar mode used in by pieces operations is enabled by the target. The TImode is not enabled on ppc. It should be checked before taking TImode to do by pieces operations. I made a patch for the generic code and testing it. With the patch, 16-byte comparison could be enabled on both ppc64 and ppc. Thanks Gui Haochen
