Is QAT multithreaded?

[Phy Sik]

Hey, just a quick question.

I'm running my QAT+ installation in a virtual machine (ubuntu 20.04 LTS in VirtualBox).

At the moment, this virtual machine has 4 cores and 8GB of RAM. 

I use complex Testlists and I read reference-values like a ddc and cross profile data , machine-parameters and so on from a big excel-file with 20+ sheets (so I don't need to mess in the code in case sth changes).  

I also work a lot on uploaded dicom-x-ray-images for constancy checks, so I use a lot of composite tests (upload a dicom-image, perform stuff on the data, put results like contrast, field size and so on in a dict and use this data in further tests...)

Long story short, would be more cores in my VM be any good or is QAT single-core-software? I realised that QAT is quiet slow in some situations, e.g. when a testlist is started and there is a hidden composite-test to read in a lot of data from said excel-file. But I dont want to waste any server resources if QAT is not able to use them...

Thanks a lot for a quick info

[tbe...@gmail.com]

Hi,

QATrack will probably not benefit from more cores. There is no automatic multiprocessing or multithreading in software. You normally have to implement that yourself in your code.

Python itself is using something that is called the global interpreter lock (GIL) which makes sure "that only one thread executes Python bytecode at a time".

Since Python 3.13 there is an option to disable the GIL but that is still experimental and the multithreading has still to be implemented by the programmer.

regards

Thomas

[Cody Crewson]

While QATrack is currently not really multi-threaded, I have been able to show if you use any parallel processing library within a composite test, that test will be multi-threaded. For some of your tests could I suggest you look into QCPump? It's not ideal but it might help for some of your tests.