I have a multithreaded file converter that I’m working on. On Windows, it puts each file that’s being converted in its own thread and uses 100% CPU (on all cores) all the time. It’s awesome! On Ubuntu, I get 100% on the first core and ~10% on all the rest. The performance is poor and disappointing.
I’m using Threads, all within a SwingWorker so I don’t freeze the GUI. I use thread.join on all threads so I perform a certain task when all threads are complete. I have not changed the code between OS’s. Is there a feasible way to fix this?
It was very dumb and I don’t quite understand why, but shortly after I posted this, I transferred all of my files to my Ubuntu partition, and it’s just as fast (if not faster) than the Windows one. Not sure why moving files would make it go faster? Perhaps my real issue was that since they were on different file systems, my bottleneck was I/O. Converting just one file from the NTFS partition took 3x longer than if I moved it to the ext4 partition. (And Yes, these are all on the same SSD)