Nice and powerful, but veeeeery slow on big files

[pawel-kowalski]

Opening a complex DWG file (or DXF, it doesn't make a difference) takes a very long period of time. For example, 7MB file is opening for 5 minutes now, and I'm still waiting... qcad-bin consumes 100% cpu (one core) and no more than 5% memory. When the file is finally opened, I can work on it, but each time I try to make print preview, add another layer, etc. I have to wait a fair bit of time.

I hope this will change in the future...

Best Regards,
Paweł Kowalski

[andrew]

Thanks for the feedback.

QCAD might not be suitable (in the near future) for your type of work. A 7MB DWG file is indeed considered to be huge for QCAD.

QCAD (unlike other CAD applications) does not work directly with the DWG format in memory. It has its own memory model which is independent of the DWG format. This is a good thing but requires conversion when loading DWG files which can be slow.

Performance will certainly improve in future releases but it is not a top priority.

[pawel-kowalski]

Thanks for your reply.

It's not a good news for me. It seems that data structures implemented in QCAD are not optimal for big projects, at least for now. After converting from DWG to QCAD internal memory model, the program works too slow.

Loading DWG files doesn't have to be instant, but I hope that in the future it's possible to speed up work on big files, which are already opened.

Best Regards,
Paweł Kowalski

[pawel-kowalski]

Is it possible to make QCAD use more than 1 cpu core? While still not optimized for big files, this could speed up work up to 3 times on quad core cpus...

Best Regards,
Pawel Kowalski

[andrew]

Unfortunately, the math is not that simple. Multiple CPUs mostly make sense whenever there are multiple resources involved (e.g. multiple files to process, multiple web pages to render, multiple processes to run, etc.). If that is not the case (e.g. one big drawing to render on one screen / multiple entities to render but into one screen buffer) things look different. Every thread ultimately waits for that one resource to become available.

We will eventually look into performance issues, but stability issues are the top priority at the moment.