Colleen Bertoni comments

Results 47 comments of


                                            Colleen Bertoni

CMake integration

Thanks for this! We tried it out on our system and it works to build more things than it used to! However, I was hoping that I could more or...

__syncthreads() warning on divergence

Thanks, I will try out compute-saniter to see if it flags anything else in the code!

MPI_File_write_at with a GPU device buffer on Intel GPUs

Thanks for the responses! So if I understand right, in theory the code is "ok" but ROMIO doesn't have it implemented so it should fail. (Or should is the code...

MPI_File_write_at with a GPU device buffer on Intel GPUs

Thanks a lot! I tried to test it out on Aurora, with ``` rm -rf mpich_7108 git clone -b romio-device-buf [email protected]:raffenet/mpich.git mpich_7108 rm -rf install_7108 mkdir install_7108 cd mpich_7108 git...

MPI_File_write_at with a GPU device buffer on Intel GPUs

Ah, it was my bad -- I forgot to `rm test` before running. #7108 fixes it indeed. @pkcoff can you check if the branch works for you?

Using malloc_shared with MPI_File_write_at_all on Intel GPUs

Thanks a lot for testing! Yes, the bad output is from Aurora. I just checked on Sunspot as well, and I realized that it does seem to work ok from...

Using malloc_shared with MPI_File_write_at_all on Intel GPUs

Oh, great! Does it need to be on top of [#7042](https://github.com/pmodels/mpich/pull/7042) (Alex's branch?) to work, or is it ok on top of main too?

Hang with MPI send/recv and pipelining on Intel GPUs

I think I can reproduce -- did the output look like the following for you? ```console + mpiexec -np 24 --ppn 12 build/tests/staggered_invert_test --dslash-type asqtad --compute-fat-long false --solve-type direct-pc --solution-type...

Hang with MPI send/recv and pipelining on Intel GPUs

It is still hanging with `mpich/opt/develop-git.6037a7a` (aurora_test branch) and `MPICH_CVAR_CH4_OFI_HMEM_ENABLE=1`. Is it worth trying main?