How to make make more CPU work in cell2location?

[Rong-ao]

Please use the template below to post a question to https://discourse.scverse.org/c/ecosytem/cell2location/.

Problem

• @vitkl Hi, thanks for your explanation! I also met the same warning as @limin321 did in The 'train_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the num_workers argument to num_workers=15 in the DataLoader` to improve performance. #407 , and I want to ignore it----but it's struggling that I don't have GPU resources for acceleration so that I want to set the parameter "num_workers" to speed up.
  or will it help in this case?

[vitkl]

num_workers parameter is not relevant due to how cell2location dataloader works

GPU acceleration is essential except for very small data, a few 10s of observations (such bulk decomposition) or a few genes. It should be possible to use cloud platforms at a reasonable cost (a few $ or $$) - especially compared to the cost of a Visium experiment.