Skip to content

[Flux] Remove Redundant Host & Device Sync #14711

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
gautham-kollu merged 3 commits into from
Sep 20, 2025
Merged

[Flux] Remove Redundant Host & Device Sync #14711

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5fd1fde
[Flux] Remove redundant host & device sync.
alpha0422 Sep 10, 2025
eaf0475
Merge branch 'main' into wkong/flux-sync-free
gautham-kollu Sep 17, 2025
8ff0664
Merge branch 'main' into wkong/flux-sync-free
gautham-kollu Sep 19, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions nemo/collections/diffusion/models/flux/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,7 +66,7 @@ def flux_data_step(dataloader_iter):
else:
_batch = batch

_batch['loss_mask'] = torch.Tensor([1.0]).cuda(non_blocking=True)
_batch['loss_mask'] = torch.ones(1, device="cuda")
return _batch


Expand Down Expand Up @@ -746,7 +746,7 @@ def _prepare_latent_image_ids(
batch_size, latent_image_id_height * latent_image_id_width, latent_image_id_channels
)

return latent_image_ids.to(device=device, dtype=dtype)
return latent_image_ids.to(device=device, dtype=dtype, non_blocking=True)

def _pack_latents(self, latents, batch_size, num_channels_latents, height, width):
# pylint: disable=C0116
Expand Down
Loading