Originally posted by: psjones

I'm using cplex_studio126.linux-x86-64 on a machine with 16 processing units available (nproc: 16) and I explicitly set the number of threads for CPLEX to 16 using the python wrapper: lp_instance.parameters.threads.set(16). However, CPELX still uses only one thread when solving a lp. I know that the way I use the parameter is correct since on my dual core laptop, I can set the thread count to either 1 (sequential) or 2, and this is reflected correctly when I run the program. But this is not the case when I use the machine with 16 processing units available and the program only uses 1 thread. I greatly appreciate any help.

Output of lscpu command:

Architecture:          x86_64

CPU op-mode(s):        32-bit, 64-bit

Byte Order:            Little Endian

CPU(s):                16

On-line CPU(s) list:   0-15

Thread(s) per core:    1

Core(s) per socket:    1

Socket(s):             16

Did you change any other parameters to non-default values? In particular, did you change the LP solve algorithm?

Could you please show us the CPLEX log output for your solve?

Originally posted by: psjones

Hi Daniel,

I haven't changed any of the default parameters and don't see any parallelism (basically when I run the top command I only see 1 of the 16 cpus has a utilization of 99% and the other 15 are idle, moreover it takes the same time to find a solution compared to when I solve the same problem on a single core machine). When I add  lp_instance.parameters.threads.set(16) (which from my understanding, should enable the opportunistic parallelism), I basically still get the same performance and run time.

I'm not quite sure how I can get the verbose log output, but attached is whatever was printed on the screen.

Thanks!