Performance Comparasons between 2 multi-core computers

dingjun_chencmgl_ca — Tue, 11 Sep 2012 16:14:44 GMT

HI, All,

I have tested a software product IMEX 2012.10 on two muti-core computers and the performance are surprise to me. Please see below:

1. the 1st Computer:

Processor: Intel(R) Xeon(R)CPU X5680 @ 3.33GHz 3.33GHz(2 processors)

Installed memory(RAM): 48.0 GB

System type: 64-bit OS

2. the 2nd computer:

Processor: Intel(R) Xeon(R)CPU X7560 @ 2.27GHz 2.26GHz(4 processors)

Installed memory(RAM): 256.0 GB

System type: 64-bit OS

The test results are given below:

Elapsed Time: 27733.59 seconds (OpenMP parallel IMEX 2012.10 running on the 2nd computer with 32 cores)

Elapsed Time: 23831.27 seconds (OpenMP parallel IMEX 2012 .10 running on the 1st computer with 12 cores)

Could someone give me some advices on how to improve the performance on the 2nd computer with 32 cores? what caused the differences? I am surprise there is such a big difference in performance between these 2 computers. Thanks inadvance.

Best regards,

Dingjun

You don't provide enough

TimP — Tue, 18 Sep 2012 13:16:10 GMT

You don't provide enough information to guess how your question might be topical on this forum. As no one has answered, I'll jump in and point out that multi-core software becomes more dependent on affinity settings as the number of CPUs increase. If your application doesn't scale beyond 2 CPUs (not an unusual situation) it will not be entirely surprising if CPU clock speed is important.

topic You don't provide enough in Intel® MPI Library

Performance Comparasons between 2 multi-core computers

You don't provide enough