Fİnding best mpi parameters for the MCMC (Markov-Chain Monte Carlo)

[Arman777]

I am using WSL2/Ubuntu 20.04 on Windows 10 to run a specific cosmological parameter estimation program called Cobaya. (It only runs on Linux and Mac)

Here are my specs.

Architecture:                    x86_64
CPU op-mode(s):                  32-bit, 64-bit
Byte Order:                      Little Endian
Address sizes:                   39 bits physical, 48 bits virtual
CPU(s):                          4
On-line CPU(s) list:             0-3
Thread(s) per core:              2
Core(s) per socket:              2
Socket(s):                       1
Vendor ID:                       GenuineIntel
CPU family:                      6
Model:                           142
Model name:                      Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz
Stepping:                        9
CPU MHz:                         2711.999
BogoMIPS:                        5423.99
Hypervisor vendor:               Microsoft
Virtualization type:             full

Cobaya uses the MCMC method for parameter estimation. I wonder what can be the best/optimum `mpirun ...` command to fully utilize my computer's specs for the MCMC task in WSL2/Ubuntu.

I have searched a bit, and choosing the correct `mpirun` arguments can impact the time it runs. Since my computer is nearly garbage for data analysis, I need to select the right ones.

Here is the page of the OpenMPI (https://www.open-mpi.org/doc/v3.0/man1/mpirun.1.php)

Normally I am using, `mpirun -n 2 cobaya-run test.yaml`. Even it seems I have 4 cores. I can only use 2. I guess to use all 4 cores. I can use `mpirun --use-hwthread-cpus cobaya-run test.yaml`.

Should I make some changes as well or just using `mpirun --use-hwthread-cpus cobaya-run test.yaml` will be enough ?

 

[Future Bruno]

In general, it is difficult to give a specific answer for this question as the optimal MPI command for parameter estimation can depend on multiple factors. However, in this specific case, using the command `mpirun --use-hwthread-cpus cobaya-run test.yaml` should be sufficient to take advantage of all 4 cores. It is also possible to specify different arguments in the mpirun command such as the number of processes and threads, but these arguments can differ depending on the specific application being used. Therefore, it is best to consult the documentation of the application in order to determine which arguments are best suited for the task at hand.