[shereez234]

Hello all,

I have a mesh thats of size 10 million elements. When i decompose it into 16 processors it works faster than for 24/32 or 48 processors. So I have decided to see if i can decompose this domain better or use 2 nodes in my workstation to reduce bandwidth and improve speed.


Here is my workstation specs:

Architecture:        x86_64
CPU op-mode(s):      32-bit, 64-bit
Byte Order:          Little Endian
CPU(s):              96
On-line CPU(s) list: 0-95
Thread(s) per core:  2
Core(s) per socket:  24
Socket(s):           2
NUMA node(s):        2
Vendor ID:           AuthenticAMD
CPU family:          23
Model:               49
Model name:          AMD EPYC 7352 24-Core Processor
Stepping:            0
CPU MHz:             3024.990
CPU max MHz:         2300.0000
CPU min MHz:         1500.0000
BogoMIPS:            4599.98
Virtualisation:      AMD-V
L1d cache:           32K
L1i cache:           32K
L2 cache:            512K
L3 cache:            16384K
NUMA node0 CPU(s):   0-23,48-71
NUMA node1 CPU(s):   24-47,72-95

and my hosts.txt looks like:

demontfortuniversity-Super-Server.1 
demontfortuniversity-Super-Server.2 
demontfortuniversity-Super-Server.3
demontfortuniversity-Super-Server.4 
demontfortuniversity-Super-Server.5 
demontfortuniversity-Super-Server.6 
demontfortuniversity-Super-Server.7 
demontfortuniversity-Super-Server.8
demontfortuniversity-Super-Server.9
demontfortuniversity-Super-Server.10
demontfortuniversity-Super-Server.11
demontfortuniversity-Super-Server.12
demontfortuniversity-Super-Server.13
demontfortuniversity-Super-Server.14
demontfortuniversity-Super-Server.15
demontfortuniversity-Super-Server.16
localhost.1 
localhost.2 
localhost.3  
localhost.4 
localhost.5 
localhost.6 
localhost.7 
localhost.8 
localhost.9
localhost.10
localhost.11
localhost.12
localhost.13
localhost.14
localhost.15
localhost.16

When i run " mpirun --hostfile hosts.txt -npersocket 16 transientSimpleFoam -parallel >log " it works. But i am not sure if this is the right option


When u run mpirun --hostfile hosts.txt -npernode 16 transientSimpleFoam -parallel >log " it fails to run giving me error message " There are not enough slots available in the system to satisfy the 0 slots
that were requested by the application:"


Couple of questions:

1) Is it possible for me to set up 2 nodes and 16 processors per node? What is the right command to do this?

2) Do I have to decompose the grid into node1 -> 16 proc and node2-> 16 proc ? or just a normal decomposition of say like simple "(8 2 2)" would be okay?
3) Would be it better if i install SLURM to do all this?