search

fury-gl / fury

FURY - Free Unified Rendering in pYthon.
https://fury.gl
Other
230 stars 163 forks source link

Memory usage after calling actor.slicer #711

Open ckolluru opened 1 year ago

ckolluru commented 1 year ago

Hi,

I have a 3D dataset (4000, 3000, 100) that I'm rendering with fury. The size of the dataset is about 1.2 GB.

When I call the actor.slicer function as: image_actor_z = actor.slicer(volume, affine), the memory usage after the call increases by approximately 6 GB.

Is this expected? I'm trying to use fury to eventually render datasets of size 50-60 GB on a workstation with 128 GB of RAM, so would this be possible? It seems not if the memory usage scales linearly with dataset size.

I tried debugging through the actor.slicer function and found that memory usage changed significantly at the following lines in actor.py. Any help would be appreciated.

Line 89: Increased by 1.2 GB vol = data Line 91: Decreased by 1.2 GB im = ImageData() Line 110: Increased by 1.1 GB vol = np.ascontiguousarray(vol) Line 138: Increased by 1.1 GB image_resliced.AutoCropOn() Line 256: Increased by 3.5 GB plane_colors.Update()

Thanks, Chaitanya

skoudoro commented 1 year ago

Thank you for the feedback @ckolluru.

We are implementing new way to handle this and improve performance. it might be present on the future release this year