Alex Brooks

> @abrooks98 Tim monitored the number of open FD with the draft PR (which from testing still results in the out-of-memory issue) and with the version built with `--with-ch4-shmmods=none` (which...

@zippylab and @colleeneb I just wanted to let you know that I am now able to run the XGC build from Renzo. I am working this week on debugging and...

@zippylab and @colleeneb I have made some progress today. I just pushed a number of changes which seem to fix the memory issues in some cases. I still have some...

> Previously I was cherry-picking `409baeb60540483e952e6c4623d326dca3a88592 ` from [[email protected]](mailto:[email protected]):zhenggb72/mpich.git , but it looks like that was merged in -- do you still recommend pulling that in to test, or can...

@colleeneb I've added two new patches, one that I think will resolve memory issues in other scenarios (I still need to do some more testing on this) and another that...

Marking this PR as ready for review. I and @zippylab have done some testing with XGC and so far see no unbound memory growth. I also ran the GPU test...

test:mpich/ch4/most test:mpich/ch3/most

Latest change was to fix an issue when setting to `disabled`. Previous tests will not be affected since it is a non-default case. ch4 tests: https://jenkins-pmrs.cels.anl.gov/view/mpich-review/job/mpich-review-ch4-ofi/4877/ ch3 tests: https://jenkins-pmrs.cels.anl.gov/view/mpich-review/job/mpich-review-ch3-tcp/2521/

test:mpich/ch4/gpu/ofi ✅ https://jenkins-pmrs.cels.anl.gov/view/mpich-review/job/mpich-review-ch4-gpu-ofi/330/ timeout in am-only is unrelated

> LGTM Thanks Hui! Please merge when you are ready