Looking for answers ... again (this time multithreading)
I have two machines:
1. 3.5 Ghz 6-Core Intel Xeon E5 with 16 GB of ram. Runs OS X 10.12.6
2. 2.8 Ghz Intel i7 860 4 core hyperthreaded. Runs Ubuntu 18.04
I use htop on both systems to monitor performance.
The conversion of pdf files to ppm files seems to take about the same time on both systems. (2-3 minutes).
However, when it gets to the automatic recognition part, Ubuntu is significantly faster. All 8 threads are running
at 100 percent. The total time is about 6 minutes, from the time I click the "Automatic" button until it is done.
On my Mac, all 6 processors seem to run anywhere from 35 to 55 percent. The total time in that case is about 20 minutes.
I know you can write code on the mac so that all processors run at 100%, because I have done it.
I am just wondering if there are some assumptions being made (like hyperthreading) that aren't true on a mac, or if there is something different
in the code that causes this discrepancy. If the code is the same for both, then it points to the perl implementation, perhaps. Or some multithreading problem with osx.
Getting an ubuntu box has been an eye opener for me. Even though the box is on 6-7 year old hardware, and the mac pro is slightly newer and supposedly faster, processing on an ubuntu box seems to outperform processing on Os X. I am just trying to understand why.
RE: Looking for answers ... again (this time multithreading) - Added by Alexis Bienvenüe 7 months ago
On my Mac, all 6 processors seem to run anywhere from 35 to 55 percent.
In case AMC did not recognize properly the number of parallel jobs you can run, you can set it by hand in Preferences/Main/System/Number of processes.
Thanks, I'll try that when I get back in the office. It is Thanksgiving week this week, and the university decided it was a good time to shut down all power to my building to change out a transformer...