I usually let this slider sit on its standard value (2) and asked myself what the benefit could be if I pushed the slider to its maximum position. This is what I found on my 4 core iMac 13,2 (27in, late 2012)
All cores are used to the max whatever the setting is, some gaps (not 100%) exist if set to 1
Processing times are almost identical when the slider is set to 2 and above
Processing times are only reduced by 1/3 with the slider set at 2 vs set at 1
It looks like DPL will use whatever resources are available, no matter what your settings are. Pushing the slider further than 3 did not shorten processing at all in my tests. I suppose (have not tested it so far) that higher positions make DPL interfere (annoy) more when I’d do other things in parallel…
These settings matter much more on 6, 8 or 12 core machines, platypus. Their importance is to help a photographer not completely freeze his/her machine during a large export. S/he should choose half as many images to export at a time as s/he has cores if there is other work to be done beyond reading an already open web page or PDF file, while waiting on export.
On my MacBook Pro 2012 the export uses all 4 (hyper) threads that the processor offers, even though I have this setting at 1. I tested with 2 or 3 in the past but found that the export was not faster and PL1 showed a tendency to crash back then.
Whether or not the system is still responsive depends much more on the OS (scheduler) and on the available RAM (swapping).
I checked the settings on my 12 core Mac Pro. With 5 set, PhotoLab used up all the processors. With 3 set, PhotoLab still used between 800 and 1200% but my Mac Pro was considerably more responsive. So I’m leaving it at 3 in parallel even on a 12 core machine. I might do some timing tests to see if there’s any difference in output speed but might not get around to it. Export time is not a really a big deal for me: responsiveness of PhotoLab to the sliders in a complex edit is.
On a laptop the heat on the power block with a lot of prim gets a bit much (to hot to touch!). So have cut my number down as well before things melt!
Thinking about it this is also the need to get greater efficiency into the program. I just spent a long time watching things being done, local adjustments, rather than them actually changing. Until by very hot power supply I hadn’t worried over the slow export, but now to be safe an even slower one after reducing the number of images precessed. So really both areas desperately need improvements.
Each one uses around 600MB while busy. So having lots of memory is not too important for PhotoLab.
Also from what you wrote I think that each XPCCore process uses four threads for heavy processing, which results in 400% CPU utilization for just one picture in parallel. With PRIME the picture is different, it seems to use more cores.
Therefore I would leave the setting at ceil(#cores/4), e. g. 1 for up to 4 cores, 2 for up to 8 cores, 3 for up to 12 cores.
GPU is not used during export, or at least it doesn’t move the needle.
Excellent test Calle. Your recommendations make sense to me as well for a balance of speed while retaining some basic use of the workstation (reading, uploading, moving other images around, browsing images).