Hello Prof. Lewis,
Sorry for replying to an old thread - I was wondering if this is still an issue now? i.e. Is it still recommended that we use intel before 2019 to use mpirun?
For some context, we’re trying to run CosmoMC (planck2018 branch) on a cluster, and spreading the job across multiple nodes seems to run significantly slower than on only one node, with little chain output. We’re currently using intelmpi/2021.5+intel-2022.0.
Any insights much appreciated!
Best,
Rayne
Sorry for replying to an old thread - I was wondering if this is still an issue now? i.e. Is it still recommended that we use intel before 2019 to use mpirun?
For some context, we’re trying to run CosmoMC (planck2018 branch) on a cluster, and spreading the job across multiple nodes seems to run significantly slower than on only one node, with little chain output. We’re currently using intelmpi/2021.5+intel-2022.0.
Any insights much appreciated!
Best,
Rayne
Statistics: Posted by Rayne Liu — March 17 2025