Jon Wright

Apologies if this is the wrong forum for a newbie question. My problem is more about how to use go assembler than python. I have gotten stuck trying to compile...

Single threaded 1D integration : performance glitch?

15

I am seeing something like 10 fps per core for our Eiger 4M images, which seems to be in line with the pyFAI-benchmark plots in the docs. For some reason...

Please use relative paths in poni and json files

5

When users take their data home they cannot process it any more because the paths to the calibration files are hardcoded to the root of the filesystem. If we were...

Add a flipping function based on header information...?

8

We often get edfimages that are mistakenly flipped during data collection. There is a header item which tells us this flipping was done. Can we add something to put them...

LZ4 from Intel's ipp ?

2

For the bitshuffle (LZ4) filter (on Intel) it seems the Intel IPP decompressor can be faster than the LZ4 which is currently bundled. The licensing situation is not entirely clear...

As mentioned in issue #138, you can get a speed up for certain datasets on large CPU's by using a static openmp schedule. In this pull request, all threads read...

It seems the openmp locks and (dynamic,1) overhead can become important for machines with large numbers of cores. For decompression, I could see some [improvements](https://gist.github.com/jonwright/9d0dfe175eb734430eb69827428b8003) using static scheduling:  Perhaps...

I started working on some code to call nvcomp for LZ4 decompression and then do the bitshuffle on a GPU. For now I only looked at decompression and the data...