ABataev added a comment. In D84767#2180433 <https://reviews.llvm.org/D84767#2180433>, @ye-luo wrote:
> In D84767#2180280 <https://reviews.llvm.org/D84767#2180280>, @ye-luo wrote: > >> This patch >> GPU activities: 96.99% 350.05ms 10 35.005ms 1.5680us 350.00ms >> [CUDA memcpy HtoD] >> before the July21 change >> GPU activities: 95.33% 20.317ms 4 5.0793ms 1.6000us 20.305ms >> [CUDA memcpy HtoD] >> Still more transfer than it should. > > @ABataev could you have a look? My July 21 compiler was built before > "[OPENMP]Fix PR46012: declare target pointer cannot be accessed in target > region." gets in. Are you talking about the `number of calls` value? The total number of calls will increase after the patch anyway, PTR_AND_OBJ adds 1 extra mem transfer for transferring translated pointer address. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D84767/new/ https://reviews.llvm.org/D84767 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits