Error code 700 on DLI course

Omniverse Isaac Sim
,
1

Isaac Sim Version

5.0.0

Operating System

Ubuntu 22.04

GPU Information

  • Model: RTX 4070 8GB VRAM
  • Driver Version: 570.86.10

Topic Description

I’m following DLI course Software-in-the-Loop Testing for Robots With OpenUSD, Isaac Sim, and ROS, but running the complete scene I get the errors below. My suspect is that it’s running out of memory, but since I can’t test it on other machines I can’t confirm: is it something to be expected for this scenario to require more than 8GB?

Error Messages

2025-08-16T19:34:22Z [153,468ms] [Error] [omni.physx.plugin] PhysX error: PhysX Internal CUDA error. Simulation cannot continue! Error code 700! , FILE /builds/omniverse/physics/physx/source/physx/src/NpScene.cpp, LINE 2989 2025-08-16T19:34:22Z [153,468ms] [Error] [omni.physx.plugin] Cuda context manager error, simulation will be stopped and new cuda context manager will be created. 2025-08-16T19:34:22Z [153,480ms] [Error] [carb.cudainterop.plugin] CUDA error 700: cudaErrorIllegalAddress - an illegal memory access was encountered) 2025-08-16T19:34:22Z [153,480ms] [Error] [carb.cudainterop.plugin] Failed to wait on external semaphore in CUDA. 2025-08-16T19:34:22Z [153,480ms] [Error] [gpu.foundation.plugin] Wait for external semaphore failed in CUDA submission in command list "Render graph command list (Render queue 0, device 0, frame submission index 2)"
2

Could you please provide some additional details? What specific part of the course are you on when the error occurs?

3

It happens when I try to run the Challenge - Multi-Robot Coordination section.
I load the Checkpoint4_completed_environment’s warehouse_ROS.usd scene and press PLAY.

The Franka robot moves in ready position, but when the nova should start moving (I assume) the following errors pop up and the simulation stops.

Screenshot from 2025-08-20 10-21-11
Screenshot from 2025-08-20 10-21-11640×400 109 KB

With the scene loaded but before starting the simulation, the memory usage is:

Screenshot from 2025-08-20 10-32-00
Screenshot from 2025-08-20 10-32-00565×169 41.5 KB

This is the full output in CLI:

cli_output.txt (319.6 KB)

I’m aware that my GPU is below the minimum requirements for Isaac SIM (8GB VRAM), and in the documentation it states that simulating sensors may require even more VRAM than what stated in the minimum requirements (16GB).

I’m also interested in the DLI courses for Isaac Lab, and in the documentation it states that Isaac Lab may require even more VRAM: do you have any suggestion on what’s the minimum requirement to run all the Isaac Lab DLI courses?
I’m referring to those in the Getting Started With Isaac Lab section of Robotics Fundamentals Learning Path.

4

I could try on a machine with 12GB VRAM but there’s exactly the same error.
Maybe it’s the Isaac Sim version? (5.0 instead of 4.2 or 4.5 cited in the course)

5

Could you provide the link of the section?

6

Here it is:
https://apps.learn.learn.nvidia.com/learning/course/course-v1:DLI+S-OV-39+V1/block-v1:DLI+S-OV-39+V1+type@sequential+block@6cd288aba4744d49a47b2217f759b764/block-v1:DLI+S-OV-39+V1+type@vertical+block@9872691f533e4703963d77925b050159

It’s the “Challenge” section of this course:

7

The url doesn’t lead to anywhere. Could you help double check it?

8

Since the course requires login maybe the link is user-specific. Can this screencast help pinpointing the right section?

I highlight the links before clicking them.

The course assets are available here:
https://dli-lms.s3.us-east-1.amazonaws.com/assets/s-ov-39-v1/DLI_SIL_online_dli.zip