No performance gains when using parallel computing (cpucounts)?

[Jesse Lam]

Hello,

When running MCCL, I do not seem to be getting any faster simulation whether using cpucount=1 or cpucount=8. I am wondering what could be the problem? I am running the default "infile_ellip_FluenceOfRhoAndZ.txt" with N=1000.

First run: cpucount=1

Time: 18.89 seconds

Second run: cpucount=8

Time: 18.49 seconds

I attempted this on an AMD Ryzen 7 4700U and Intel Xeon Silver 4216 with no luck.

Thank you!

Kind regards,

Jesse Lam

[Lisa Malenfant]

Hi Jesse,

Could you make sure you are using the latest version of MCCL, the cpucount was added in version 5.1 and the latest version is 6.2, here is the link to download the latest:
https://github.com/VirtualPhotonics/Vts.MonteCarlo/releases/tag/Version_6.2

Let us know if you need further assistance running the parallel version.

Thank you,

Lisa

[Jesse Lam]

That resolved my issue, thank you! The download link at https://virtualphotonics.org/software linked to version 4.9, which I mistakenly assumed was the latest version.

[Lisa Malenfant]

Hi Jesse,

I'm glad that solved your issue. Thank you for alerting us to that invalid link, I have updated it on our website. We separated out the MCCL application and it looks like some of the links were not updated.

Lisa