To those who have experience on Linda,
As I ran a medium-size job in single processor with requested
memory=13mw, it could successfully ran even if 2 or 3 other jobs were
running meanwhile. To do multi-processor calculation, I increased
Nproclinda=3 ( we have four nodes, each with 2CPU) and kept the same memory
request or less (8mw), initially the job really went to other nodes, however
after a while it crashed regardless of jobs running or idle in other nodes.
The error message read like this,
Linda panic on node master.cl.usc.edu: No more memory for Tuple Space
forrtl: error (76): IOT trap signal
Fatal: ran out of heap space
ntsnet: master process exited with return value 1
Thanks for your help.
Department of Chemical Engineering
University of South Carolina
Columbia, SC 29208