Closed rezib closed 6 years ago
This feature would be more usefull if #10 is also developed.
This one should be considered low priority, it should not be that useful indeed if platform description files are well defined.
I think it is not relevant any more given the possible of using -w
to specify the number of nodes. For example: ubench run -b hpcc -p zbook15 -w 2 4 8
When a user run a benchmark, it could use quite a lot of cores/nodes. It should be possible to limit the number of cores/nodes so that all runs above those limits are discarded. Eg: