Trackmate spot detection very slow

Hi all,
Very naive question here. I am using trackmate to detect spots from single molecule TIRF data. As there are a large number of spots I have switched to a windows powered analysis machine in an attempt to process the data faster. However, despite having 28 cores and 56 threads the detection is taking orders of magnitude longer than my macbook pro, around 6 mins compared to 7 seconds respectively. Not being a frequent windows user, I wonder if there is some issue I am not aware of with a recent update to trackmate etc? The CPU usage is only 3% and shows all 56 threads in the log trackmate so everything seems to be working there.
Any advice / direction to some simple checks would be much appreciated. Apologies again for the very basic question. If you need any more details let me know.

Cheers!

Hello Mike

This is unexpected. Maybe you did not configure Fiji to use all your cores?
What does Edit > Options > Memory & Threads… show?

Hi Jean-Yves,
Currently it shows:
Maximum memory: 146111 MB
Parallel threads 56

Thanks,
Mike

Looks like a cool PC :slight_smile:

I don’t know what is happening then. TrackMate should use the number of threads you specify there. Can you try to download a fresh Fiji on this machine to check whether this is linked to Fiji configuration?

It was supposed to be! We specced it out for this kind of work but doesn’t seem to be happy about it! Yeah, I re-installed and made sure everything was up to date but still no luck. I also have linux on the PC, yet to try that… Failing that ill get my hands on a windows laptop and compare everything and see if there is just something simple I have missed.

Just to be sure: did you check you’re using the correct pixel spacing and matching parameters for the spot diameter? I’ve seen longish processing times in TrackMate when the spot detection parameters don’t match the calibration of the input image.

1 Like