Richard Brown

Ok, I now understand the code won't work for the subject-based data. > My code (prepare_data.ipynb) does not calculate any centroids. Although you don't calculate any centroids in `prepare_data.ipynb`, the...

The code snippet mentioned [here](https://github.com/anjany/verse/issues/9#issue-1068340861) should highlight what I mean. I think that you are saying that the functionality works, but when using the subject-based data it will give different...

We really need to make progress on [this PR](https://github.com/Project-MONAI/MONAI/pull/3956) before adding in point transforms (which is definitely on our agenda).

`cache_rate` is 1.0 by default. Have you experimented with lower fractions to see if that has an impact? I wonder if the total memory of your loaded dataset is so...

~41GB on disk or with the transforms applied? From the screenshot of your htop of server B, I wondered if it was because you were running out of memory. Perhaps...

Let's not complicate things with `SmartCacheDataset` just yet. I'm not sure how you've parameterised your `SmartCacheDataset`, but for example if `replace_rate=1`, all data is going to be refreshed every time,...

We should be able to rule the GPU out fairly easily by modifying the code to not copy to the GPU. i.e., just: ```python transforms = Compose(some_transforms) ds = CacheDataset(big_data,...

The following code: ```python import time from monai.data import CacheDataset, DataLoader from tests.utils import create_test_image_3d, make_nifti_image from monai.transforms import Compose, LoadImaged, AddChanneld, ToTensord def main(): im_shape = 100 num_ims =...

Any movement on this one?

> This can currently be done by inspecting the ITK modules variables. Here is what I did in SimpleITK: Do you use this for IO? I gave it a go...