Skip to content

[MGPU] App: pin Kit renderer to single GPU under ISAACLAB_PIN_KIT_GPU #5933

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

Draft
hujc7 wants to merge 1 commit into
base: develop
Choose a base branch
from
+33 −0
Draft

[MGPU] App: pin Kit renderer to single GPU under ISAACLAB_PIN_KIT_GPU #5933

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
14 changes: 14 additions & 0 deletions source/isaaclab/changelog.d/jichuanh-mgpu-pin-kit-resources.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
Added
^^^^^

* Added ``ISAACLAB_PIN_KIT_GPU`` env var for :class:`~isaaclab.app.AppLauncher`.
When set to a truthy value, appends ``--/renderer/multiGpu/enabled=False``,
``--/renderer/multiGpu/autoEnable=False`` and ``--/renderer/multiGpu/maxGpuCount=1``
to the Kit command line so each Kit process touches only its assigned
GPU (rather than enumerating every visible GPU at startup). Used by the
multi-GPU CI workflow to prevent the shared cubric / PhysX-fabric
GPU-interop context across sibling shards that surfaces as
``[Error] [omni.physx.plugin] Stage X already attached`` and
``SimulationApp.close`` hangs (see https://github.com/isaac-sim/IsaacLab/issues/3475
and NVBug 5687364). Off by default; single-GPU and user-facing rendering
paths are unchanged.
19 changes: 19 additions & 0 deletions source/isaaclab/isaaclab/app/app_launcher.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1060,6 +1060,25 @@ def _resolve_device_settings(self, launcher_args: dict):
launcher_args["physics_gpu"] = self.device_id
launcher_args["active_gpu"] = self.device_id

# Pin Kit's renderer to a single GPU when ``ISAACLAB_PIN_KIT_GPU`` is
# truthy. The default ``apps/isaaclab.python.headless.kit`` sets
# ``renderer.multiGpu.enabled = true`` + ``renderer.multiGpu.autoEnable
# = true``, so each Kit process enumerates every visible GPU at
# startup. Under concurrent multi-GPU CI shards (``--gpus all`` per
# container, one Kit per non-default cuda device), that produces a
# shared cubric / PhysX-fabric GPU-interop context across sibling
# processes -- surfacing as ``[Error] [omni.physx.plugin] Stage X
# already attached`` mid-test and ``SimulationApp.close`` hanging
# >52s in teardown (see https://github.com/isaac-sim/IsaacLab/issues/3475
# and NVBug 5687364). Kelly Guo's documented WAR (#omni-kit thread,
# 2024-2025): set ``renderer.multiGpu.enabled = false`` + ``maxGpuCount
# = 1`` so each Kit only touches its assigned GPU.
if os.environ.get("ISAACLAB_PIN_KIT_GPU", "0").lower() not in {"", "0", "false", "no", "off"}:
sys.argv.append("--/renderer/multiGpu/enabled=False")
sys.argv.append("--/renderer/multiGpu/autoEnable=False")
sys.argv.append("--/renderer/multiGpu/maxGpuCount=1")
logger.info("ISAACLAB_PIN_KIT_GPU enabled: pinning Kit renderer to a single GPU")

# Defer importing torch until after SimulationApp starts. Importing
# torch can import NumPy/OpenBLAS, whose at-fork handlers can crash
# Kit's platform-info fork during startup.
Expand Down
Loading