[ENH] add gpu tracking to pyAFQ #962
Conversation
|
@arokem this is ready for review |
There was a problem hiding this comment.
This is very cool. I think that it would be nice to make the API of the new "gpu_track" function as similar as possible to the API of "track", so that they can more readily be interchanged (and so that they can share much of the boiler-plate code). So, for example, the seed mask is generated within "track", but it is assumed that it is passed into "gpu_track" (and always as Nifti1Image!). The new "gpu_track" function needs data and gradient table, whereas "track" needs model params. WDYT?
| # Modified from https://github.com/dipy/GPUStreamlines/blob/master/run_dipy_gpu.py | ||
| def gpu_track(data, gtab, seed_img, stop_img, | ||
| seed_threshold, stop_threshold, thresholds_as_percentages, | ||
| max_angle, step_size, sampling_density, ngpus): |
|
I agree that we should make the gpu_track API as similar to the track API as possible, but I think we should do that in a future PR, once we have used gpu_track a little more |
|
OK - that makes sense. Other than that, I think that this is ready to merge, so if you agree, please go ahead! |
No description provided.