Skip to content
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

Move SystemMonitor's GPU initialization back to constructor #4866

Merged
merged 6 commits into from
Jun 3, 2021
Merged

Move SystemMonitor's GPU initialization back to constructor #4866

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
d860e58
Always use index 0 to get NVML GPU handle
pentschev Jun 1, 2021
80224a8
Move SystemMonitor's GPU initialization back to constructor
pentschev Jun 1, 2021
ddf9a43
Use nvmlDeviceGetHandleByIndex directly
pentschev Jun 1, 2021
79b315b
Remove redundant nvmlInit check
pentschev Jun 1, 2021
8218605
Revert "Use nvmlDeviceGetHandleByIndex directly"
pentschev Jun 2, 2021
95d93f4
Revert "Always use index 0 to get NVML GPU handle"
pentschev Jun 2, 2021
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
12 changes: 4 additions & 8 deletions distributed/system_monitor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,8 +40,9 @@ def __init__(self, n=10000):
self.quantities["num_fds"] = self.num_fds

if nvml is not None:
self.gpu_name = None
self.gpu_memory_total = None
gpu_extra = nvml.one_time()
self.gpu_name = gpu_extra["name"]
self.gpu_memory_total = gpu_extra["memory-total"]
self.gpu_utilization = deque(maxlen=n)
self.gpu_memory_used = deque(maxlen=n)
self.quantities["gpu_utilization"] = self.gpu_utilization
Expand Down Expand Up @@ -90,12 +91,7 @@ def update(self):
self.num_fds.append(num_fds)
result["num_fds"] = num_fds

# give external modules (like dask-cuda) a chance to initialize CUDA context
if nvml is not None and nvml.nvmlInit is not None:
if self.gpu_name is None:
gpu_extra = nvml.one_time()
self.gpu_name = gpu_extra["name"]
self.gpu_memory_total = gpu_extra["memory-total"]
if nvml is not None:
gpu_metrics = nvml.real_time()
self.gpu_utilization.append(gpu_metrics["utilization"])
self.gpu_memory_used.append(gpu_metrics["memory-used"])
Expand Down