Leginon

DM does the integration and pass the image to Leginon. It may or may not be faster, but K2 camera is powerful for its own reason. Berkeley is doing what you propose already, but we are not sure if it is the cause of their failure to save movie frames in super-resolution mode which is an operation of DM, not Leginon.

Can Leginon take advantage of a multicore machine to speed up computations on the leginon side?
Would it be possible for instance to decouple the display of acquired data and the procedures running for the next hole in line?

Amedee

First question: Leginon runs on single python instance and therefore limits it to single core. It should in principle speed up by taking advantage of a multicore machine. It is something we have meant to work on but never finished up.

Second question: The display can be turned off, and you do get a good speed up for doing so on big images. I would recommend you do that when you no longer need to look at them right there. Use web imageviewer instead. We did write a special option to move the stage while the camera is reading out. In other words, the stage could move while K2 is assembling the frames. However, it was only useful in case of Plastic embedded Serial Section imaging where they spend most time moving stage all over the grid and thus takes longer. Since the camera can not be used for next imaging as far as we know until the frames are saved, the time saved by this is not significant for cryo-em.

Thanks! I hope you'll manage to implement the multicore soon! :-)

How do we turn off the image display? I'll do that asap!

Any Acquisition Node (icon looks like a camera) /Settings/Advanced > uncheck "Display Image"