Skip to content

Fix Ray memory issue caused by in-flight inference results #39

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

Open
CarlCypress wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Fix Ray memory issue caused by in-flight inference results #39

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
da0662a
chore: ignore .idea
CarlCypress Dec 15, 2025
eec2a99
fix: Ray in-flight results to reduce memory usage
CarlCypress Dec 15, 2025
2ffc44d
fix: ignore .idea
CarlCypress Dec 15, 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
3 changes: 2 additions & 1 deletion .gitignore
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,4 +48,5 @@ cufile.log

# annotation tool
/annotation_tool/application
/annotation_tool/Dockerfile
/annotation_tool/Dockerfile
.idea/
50 changes: 42 additions & 8 deletions cellvit/inference/inference_memory.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,6 +141,8 @@ def process_wsi(
]

call_ids = []
MAX_IN_FLIGHT = 4
inference_results = []

self.logger.info("Extracting cells using CellViT...")
with torch.no_grad():
Expand All @@ -158,15 +160,39 @@ def process_wsi(
else:
predictions = self.model.forward(patches, retrieve_tokens=True)
predictions = self.apply_softmax_reorder(predictions)
call_id = batch_actor.convert_batch_to_graph_nodes.remote(
predictions, metadata
)
# ==================== {origin code start} ====================
# call_id = batch_actor.convert_batch_to_graph_nodes.remote(
# predictions, metadata
# )
# call_ids.append(call_id)
# pbar.update(1)
# pbar.total = len(wsi_inference_dataloader)
# ==================== {origin code end} ====================
# ==================== {fix code start} ====================
call_id = batch_actor.convert_batch_to_graph_nodes.remote(predictions, metadata)
call_ids.append(call_id)
if len(call_ids) >= MAX_IN_FLIGHT:
ready_ids, call_ids = ray.wait(call_ids, num_returns=1, timeout=None)
ready_results = ray.get(ready_ids)
inference_results.extend(ready_results)
ray.internal.free(ready_ids)
del ready_results, ready_ids
# ==================== {fix code end} ====================
pbar.update(1)
pbar.total = len(wsi_inference_dataloader)

self.logger.info("Waiting for final batches to be processed...")
inference_results = [ray.get(call_id) for call_id in call_ids]
# ==================== {origin code start} ====================
# inference_results = [ray.get(call_id) for call_id in call_ids]
# ==================== {origin code end} ====================
# ==================== {fix code start} ====================
while call_ids:
ready_ids, call_ids = ray.wait(call_ids, num_returns=min(8, len(call_ids)), timeout=None)
ready_results = ray.get(ready_ids)
inference_results.extend(ready_results)
ray.internal.free(ready_ids)
del ready_results, ready_ids
# ==================== {fix code end} ====================
del pbar
[ray.kill(batch_actor) for batch_actor in batch_pooling_actors]

Expand All @@ -191,10 +217,18 @@ def process_wsi(
batch_cell_tokens,
batch_cell_positions,
) = batch_results
cell_dict_wsi = cell_dict_wsi + batch_complete_dict
cell_dict_detection = cell_dict_detection + batch_detection
graph_data["cell_tokens"] = graph_data["cell_tokens"] + batch_cell_tokens
graph_data["positions"] = graph_data["positions"] + batch_cell_positions
# ==================== {origin code start} ====================
# cell_dict_wsi = cell_dict_wsi + batch_complete_dict
# cell_dict_detection = cell_dict_detection + batch_detection
# graph_data["cell_tokens"] = graph_data["cell_tokens"] + batch_cell_tokens
# graph_data["positions"] = graph_data["positions"] + batch_cell_positions
# ==================== {origin code end} ====================
# ==================== {fix code start} ====================
cell_dict_wsi.extend(batch_complete_dict)
cell_dict_detection.extend(batch_detection)
graph_data["cell_tokens"].extend(batch_cell_tokens)
graph_data["positions"].extend(batch_cell_positions)
# ==================== {fix code end} ====================

# cleaning overlapping cells
if len(cell_dict_wsi) == 0:
Expand Down