OpenMP builds, any difference when running fluid sims?

I was suggested to me by someone in this forum to install an OpenMP build of Blender to take advantage of all my cores, so I installed this one:

That’s a zip file with the contents of the Blender folder inside c:\program files, so I moved the original folder to a backup location and copied the new files and folders. Other than a few differences in the GUI, I don’t seem to find it that different when it comes to running fluid sims. I mean, the official build runs the sim on all cores, at least according to what Task Manager shows me. The higher the resolution, the more CPU it uses, but it’s definitely running on all cores.

So is there any point in installing a separate build with OpenMP?

The official blender build already have openmp enabled so there is no point but you could try one of the experimental VS2013 builds on and see if it runs faster (if you use windows).

Sorry, I don’t know all the lingo, what is VS2013? Is it supposed to make fluid simulation faster? I ggogled blender vs2013 and I saw a lot of people talking about it, but nothing that tells me what is it and how is it different from the official build. Then on that page you linked there is also a MinGW build, which I don’t know what it is.

MinGW and VS2013 are just different compilers… pretty much to make somethign that runs on your computer, you need to convert human readable code into computer code… compilers do this… different compilers optimize human code in different ways which alllows for some to be faster.

the MinGW build on is quite old… just try the VS2013 version (VS2013 means visual studio 2013… the official compiler from microsoft… normal builds are VS2008 i believe)

Oh, I see. I’m running the VS2013 build now, on the plus side the water simulation has all the cores running at 100% most of the time, except every now and then when it comes down and goes up right away. However, I don’t see the progress bar going much faster than before, but of course my perception is not a good test. I assume that if all the cores are running at 100% whereas with the official build they were jumping up and down between 30 and 70, it has to be faster.