For what it's worth, I ran into the same issue on the same platform (ms windows) about 2 years ago. Do note however that this was using my own tesseract build!
My investigation then showed OpenMP, once triggered to start, would run my 16 cores at 100% forever, irrespective of any work done. Using a sampling profiler I found OpenMP was simply running 16 threads were every thread that wasn't given any work at that moment was idling by spinning, checking the work queue, thus running the CPU to max temp without being very useful. This experience with OpenMP at the time was in line with what I observed OpenMP doing with other test applications on my machine (not built by me): CPU nice & quiet until #openmp pragma is hit, then *BAM!* CPU maximg out all cores until end of application. The stuff that can run multithreaded does, but that's always only part of the code / run-time, but OpenMP kept my cores at max throttle by spinning during the intermissions, until the application is terminated... so the preliminary conclusion was it was an issue inside OpenMP (or me missing non-obvious setting XYZ for OpenMP). Anyway, I booted OpenMP off my system and went back to doing multi threading old skool, which is sometimes hard but always felt more comfortable to me. Take-away: if you want to investigate what happens over at yours, grab a sampling profiler (I used a commercial one from Intel at the time IIRC) and build from C/C++ source (or other means to get legible function names from debug info in the profiler run reports), e.g. using Visual Studio. Its work, its effort, but nobody else can look into your box(es) so you'll otherwise always depend on others' guesswork. On Sat, 25 May 2024, 12:54 Kassim Papa, <kassimp...@gmail.com> wrote: > Yeah sure, > > We just never were able to get it to work no matter what. But now we have > some lead. We'll try replicating it with the tesseract executable. > But for example just knowing that openmp is disabled in tesseract was > huge. And I'd like charlessw to answer us on that front. > > Le samedi 25 mai 2024 à 12:50:23 UTC+2, zdenop a écrit : > >> Well, I would suggest making a replicable case that prove the problem if >> you want the help. >> Based on the description you provided nobody can help you (neither >> charlessw) >> The problem could be somewhere in your code, in C#, in the tesseract, or >> even in your environment/OS... >> You observed the problem => you need to narrow down where is the source >> of the problem. >> >> >> Zdenko >> >> >> so 25. 5. 2024 o 12:28 Kassim Papa <kassi...@gmail.com> napísal(a): >> >>> I do not claim anything. >>> >>> Thank you for your proposition. We will test that and post on their >>> github (charlessw, the guy who made the wrapper) >>> >>> Stephan weil closed the issue on the github of tesseract saying : >>> >>> "The Tesseract for Windows which is provided by UB Mannheim does not >>> have this issue: it runs always single-threaded because it was built with >>> OpenMP disabled. You did not say what Tesseract binary and which version >>> you used." >>> >>> So I guess you must be right, this is where our effort should go. >>> >>> But I didn't even know that, I don't understand all those openmp changes >>> that have been made. COuld you explain them to me? Since the issue is >>> closed I cannot talk to stephan weil anymore. >>> >>> >>> Le samedi 25 mai 2024 à 12:12:21 UTC+2, zdenop a écrit : >>> >>>> You need to replicate it with the tesseract executable if you want to >>>> claim it is Tesseract problem.... >>>> >>>> Zdenko >>>> >>>> >>>> so 25. 5. 2024 o 12:05 Kassim Papa <kassi...@gmail.com> napísal(a): >>>> >>>>> We use a C# wrapper. >>>>> This .net library found on nugget : >>>>> https://github.com/charlesw/tesseract >>>>> >>>>> >>>>> >>>>> Le samedi 25 mai 2024 à 11:14:36 UTC+2, zdenop a écrit : >>>>> >>>>>> How did you install tesseract? >>>>>> >>>>>> What is the output of `tesseract -v`? >>>>>> >>>>>> >>>>>> Zdenko >>>>>> >>>>>> >>>>>> so 25. 5. 2024 o 11:03 Kassim Papa <kassi...@gmail.com> napísal(a): >>>>>> >>>>>>> Current Behavior : >>>>>>> >>>>>>> Despite putting omp_thread_limit=1 tesseract still use all cores on >>>>>>> my machine (i7-7th - windows 10). >>>>>>> >>>>>>> We used this (at the beginning of the code) : >>>>>>> Environment.SetEnvironmentVariable("OMP_THREAD_LIMIT", "1"); >>>>>>> >>>>>>> And this ( a batch) : >>>>>>> @echo off >>>>>>> set OMP_THREAD_LIMIT=1 >>>>>>> start "" "path_to_your_application.exe >>>>>>> >>>>>>> We have 1 big image Tesseract takes 7 second when we go over it at >>>>>>> once. >>>>>>> >>>>>>> When we divide the image in 4 and run 4 instances of tesseract in >>>>>>> parallel it take 7 second too : no changes at all. >>>>>>> >>>>>>> Expected Behavior : >>>>>>> >>>>>>> We should see in the task manager that tesseract only use 1 cores >>>>>>> >>>>>>> there should be a significant improvement when running 4 images in >>>>>>> parallel. Multiple people had success with this method. >>>>>>> >>>>>>> -- >>>>>>> You received this message because you are subscribed to the Google >>>>>>> Groups "tesseract-ocr" group. >>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>> send an email to tesseract-oc...@googlegroups.com. >>>>>>> To view this discussion on the web visit >>>>>>> https://groups.google.com/d/msgid/tesseract-ocr/008e9795-877c-4638-af08-0dc7e3af00ecn%40googlegroups.com >>>>>>> <https://groups.google.com/d/msgid/tesseract-ocr/008e9795-877c-4638-af08-0dc7e3af00ecn%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>>>> . >>>>>>> >>>>>> -- >>>>> You received this message because you are subscribed to the Google >>>>> Groups "tesseract-ocr" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to tesseract-oc...@googlegroups.com. >>>>> >>>> To view this discussion on the web visit >>>>> https://groups.google.com/d/msgid/tesseract-ocr/2ca978b6-0061-425a-aa2e-a7065fa5acc4n%40googlegroups.com >>>>> <https://groups.google.com/d/msgid/tesseract-ocr/2ca978b6-0061-425a-aa2e-a7065fa5acc4n%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>> . >>>>> >>>> -- >>> You received this message because you are subscribed to the Google >>> Groups "tesseract-ocr" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to tesseract-oc...@googlegroups.com. >>> >> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/tesseract-ocr/6479e4ae-c496-4061-971c-c51c15999325n%40googlegroups.com >>> <https://groups.google.com/d/msgid/tesseract-ocr/6479e4ae-c496-4061-971c-c51c15999325n%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- > You received this message because you are subscribed to the Google Groups > "tesseract-ocr" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to tesseract-ocr+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/tesseract-ocr/5dccdafb-4cf6-4170-bf6b-709a422246b5n%40googlegroups.com > <https://groups.google.com/d/msgid/tesseract-ocr/5dccdafb-4cf6-4170-bf6b-709a422246b5n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-ocr+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CAFP60foAZhs%3DxOVTAZ3KEUO6_Ru-8G%3DMOkBRSw4GW%3D_%3DtYEcvg%40mail.gmail.com.
