I'm noticing a similar (10%) performance hit after creating an exe-file, and its definitely not related to the initial startup time. I posted a new thread here over the weekend but it hasn't been accepted yet. In short, my program's runtime is about 1m31s in source/script mode, but upon compiling it with pyinstaller's one-file (-F), the exact same program takes over 14 minutes. I can confirm my program's code actually begins working within a few seconds of starting (with a debug print in the main routine). I'm more than happy to share the program's code and steps to reproduce, but I'll wait a little longer for my original post to go through instead of duplicating it here.
Cheers, MHL On Monday, June 1, 2015 at 10:19:42 AM UTC-5, davecortesi wrote: > > > The total runtime of my program as a script (opencv, numpy, matplotlib) > is about 4 to 5 seconds. As exe-file it takes about 30 to 40 > seconds to finish. Well, this in unacceptable. > > Indeed! But can you tell where the time is being spent? That is, is it 25 > seconds setting up before your code begins to execute? Or is it setting up > quickly but your code runs 10x slower? > > As described in ( > http://pythonhosted.org/PyInstaller/#how-the-one-file-program-works) the > one-file setup basically unpacks an archive to create the one-folder > organization in a temp location. So try it as a one-folder bundle. If that > eliminates the delay, it points to the one-file setup. If one-folder is > still slow, then stick in some print statements to find out, in a coarse > way, where is the major delay: Before your code starts, or during it, or > after it completes? > -- You received this message because you are subscribed to the Google Groups "PyInstaller" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/pyinstaller. For more options, visit https://groups.google.com/d/optout.
