*For example currently DENIS@Home is much faster on Mac, POGS is faster on Mac and possibly Linux and project applications at World Community Grid and elsewhere based on the molecular docking program VINA are almost always substantially faster run on Linux.
Wow! I just got a virtual machine setup running Ubuntu 16.04.1, the speed increase on VINA tasks is amazing!
I calculated some averages from the results saved at WCG, unfortunately there were only seven Windows Smash Childhood Cancer results still available compared to the ~100 Linux SCC results:WCG Smash Childhood Cancer
Average CPU time Windows 7 x64: 38.3 minutes (1h5m - 16m)
Average elapsed time Windows 7 x64: 43 minutes (1h13m - 18m)
Average CPU time VirtualBox Ubuntu: 25.9 minutes (52m - 10.8m)
(I didn't calculate the average elapsed time for SCC under Ubuntu because in nearly all cases it was identical to the CPU time)
I also had 3 task results of MCM from both Windows and Linux, small sample sizes, but here are the results:Mapping Cancer Markers
Average CPU time Windows 7 x64: 4h22m (4h25m - 4h21m)
Average elapsed time Windows 7 x64: 4h45m (4h54 - 4h35m)
Average CPU time VirtualBox Ubuntu: 4h21m (4h22m - 4h19m)
Average elapsed time VirtualBox Ubuntu: 4h25m (4h28m - 4h23m)
So it seems that SCC runs much faster under the virtual machine, while MCM might
run a bit faster. Is MCM VINA based as well?
I quite often have to reduce the CPU cores to 3/4 while running in the virtual machine to reduce lag while im using the computer, but even so, it's worthwhile for the SCC tasks.
I'll run some more SCC on Windows 7 to increase my sample size, but I remember a lot of tasks running for ~1 hour.
I really need to make a script to automate this...
edit: I also tried to run Linux Puppy, CentOS and OpenSUSE in VirtualBox, but Ubuntu was the only one I got to work with BOINC
edit2: CPU is Q6600 in case anyone is interested
edit3: 'Estimated computation size' is 14,011 GFLOPs in both Windows 7 and Ubuntu, so they seem to be the same tasks? They just run faster under Ubuntu...