These are chat archives for thunder-project/thunder

4th
May 2016
Kyle
@kr-hansen
May 04 2016 17:19
@jwittenbach @freeman-lab I had some input regarding the conversation above with @techchrj for the default for the partition argument. I have been working with multiplane tiff files. I would anticipate that most any time when loading a multiplane tiff, the default parallelism would be more effective.
Also, I have been learning that getting away from multiplane tiffs is a good idea, at least for the size of datasets I'm dealing with. It is a simple setting change on the camera to no longer output multiplane tiffs but individual frames, and is much quicker to load into Spark/Thunder (8x faster). This makes sense to me now, but for whatever reason I didn't consider how much slower the multiplane tiffs would be and just kept using them since that was the general standard for my lab.
Kyle
@kr-hansen
May 04 2016 17:26
I'd be interested to know how many labs typically use multiplane tiffs for data collection as their standard. I'd think quite a few because of their ease of use in ImageJ. Perhaps this might be useful to mention in the FAQ portion of the documentation, or perhaps a "Useful Tips" section for people like me who are learning elements of cluster computing on the fly while working with this?