MATLAB 2017a was running properly, but suddenly became very slow

7 views (last 30 days)
Hello,
I am running a big code in MATLAB 2017a in windows 10 to perform some data assimilation. The code calls cygwin to run a software written in C, the result of this software is analyzed and processed. My code was running as expected but it suddenly became very slow for some reason. I uninstalled and reinstalled MATLAB which was of no use. Even the boot speed is slow. I ran the 'bench' code and got:
bench
0.1788 0.1172 0.2570 0.0989 2.2075 1.9787
System specs are i7 quad core, 16gb RAM, 2.60GHz. If anyone has an idea that could help solve the problem, please let me know. Thanks!
  4 Comments
Jan
Jan on 9 Aug 2018
"Even the boot speed is slow" - this means, that it is not a Matlab problem and the question is off-topic in this forum. Maybe a problem with overheating of the CPU or a trojan which let the CPU search for bitcoins?

Sign in to comment.

Answers (2)

Philip Borghesani
Philip Borghesani on 8 Jun 2017
Completely outlandish idea. You could check if Windows enabled the fault tolerant heap for MATLAB. You can check by looking for this registry entry: Go to the HKEY_LOCAL_MACHINE and HKEY_CURRENT_USER versions of Software\Microsoft\Windows NT\CurrentVersion\AppCompatFlags\Layers\matlab.exe and delete the Fault­Tolerant­Heap entry. Information came from this stack overflow question.
  3 Comments
Philip Borghesani
Philip Borghesani on 8 Jun 2017
That may be something else but it could still have caused the problem, no recent MATLAB should need any AppCompatFlags to work properly on Windows 10. I would go ahead and delete any keys with matlab.exe in them.
Abhishek Subramanian
Abhishek Subramanian on 12 Jun 2017
Thank you for your answer. I am unable to download 2017a at the moment to try it out. Hopefully this suggestion will be helpful for others who might face a similar issue in the future.

Sign in to comment.


Don Zheng
Don Zheng on 8 Jun 2017
The bench outputs look normal. You might want to profile your code and first identify which section of the code is running slow.
  2 Comments
Abhishek Subramanian
Abhishek Subramanian on 8 Jun 2017
Bench test in MATLAB 2016b gave:
0.1895 0.1146 0.0582 0.0968 0.5144 0.5662
indicating that the performance is much better than what was seen in MATLAB 2017a
Jan
Jan on 9 Aug 2018
You cannot compare the benchmark output between different Matlab versions, because the code of bench.m might have changed. Care for using the same function to be sure: Create a copy of bench.m and call it bench2016b.m and run it in both versions.

Sign in to comment.

Categories

Find more on Startup and Shutdown in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!