Multigrid clean up phase, and fix shell variables for libfabric bug work around on Frontier

This commit is contained in:
Peter Boyle
2026-09-08 15:52:15 -04:00
parent a07545adc4
commit e8a82fa683
5 changed files with 62 additions and 12 deletions
+5
View File
@@ -12,3 +12,8 @@ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CRAY_LD_LIBRARY_PATH
module load emacs
export FI_MR_CACHE_MONITOR=kdreg2 # REQUIRED for device-buffer MPI on Slingshot: libfabric memhooks monitor (default) is defective, see systems/WorkArounds.txt (libfabric #11451)
# Standing avoidance of the CXI NO_TRANSLATION fault on persistent device
# buffers under hipMalloc/hipFree churn (libfabric issue #12775; ruling
# 2026-09-07).
export FI_HMEM_ROCR_USE_DMABUF=0