Open xmfcx opened 1 month ago
@msz-rai do you have any experience running the https://github.com/RobotecAI/RobotecGPULidar on WSL2?
@sasa42 could you also share the nvidia-smi
results with us?
don’t install the nvidia driver, there’s no real nvidia gpu available in WSL2.
my output:
mfc@mfc-leo:~$ nvidia-smi
Thu Jul 11 13:32:27 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 545.29.06 Driver Version: 545.29.06 CUDA Version: 12.3 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA GeForce RTX 3080 Off | 00000000:2D:00.0 On | N/A |
| 53% 75C P2 148W / 320W | 3447MiB / 10240MiB | 23% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 1583 G /usr/lib/xorg/Xorg 838MiB |
| 0 N/A N/A 5867 G ...yOnDemand --variations-seed-version 42MiB |
| 0 N/A N/A 6345 G ...,262144 --variations-seed-version=1 226MiB |
| 0 N/A N/A 1131802 G ...erProcess --variations-seed-version 23MiB |
| 0 N/A N/A 1164733 G ...erProcess --variations-seed-version 65MiB |
| 0 N/A N/A 2589250 C+G ./awsim_labs_v1.1.1/awsim_labs.x86_64 2198MiB |
| 0 N/A N/A 3169233 G ...seed-version=20240708-050118.992000 15MiB |
+---------------------------------------------------------------------------------------+
this is the Unity PlayerLog....
@sasa42 You shouldn't source Ros2 when launching AWSIM. That's the reason why lidars are not working.
Please comment out sourcing lines from the .bashrc if you've added there.
Like this:
...uh. What dou yo mean with source ros2.
I thsi case i only started ...v1.1.1/ ./awsim_labs.x86_64
I deleted the log files several times and redid the awsim start.
The errors you've posted says that you've sourced Ros2.
Please check your .bashrc file for this line:
source /opt/ros/humble/setup.bash
and comment it out.
What this line does is when you open a fresh terminal it sources ros2 right away. With this disabled you will have to do it manually when you need it.
Also please launch the awsim from a fresh separate terminal without sourcing the ros2.
.bashrc file is located at the home directory.
you can open it with:
sudo apt install gedit
gedit ~/.bashrc
And modify the file.
But you have other Nvidia Optix related errors, this still won't work.
Your only option is to dual boot with Ubuntu, run both AWSIM Labs and Autoware in native ubuntu 22.04
... this is the log after #source.... and restart in a new terminal.
needed configs are...
Autoware WSL - AWSIM WIN 1PC Autoware WSL - AWSIM WSL 1PC
Autoware WSL - AWSIM WIN 2PC Autoware Native - AWSIM WIN 2PC Autoware Native - AWSIM MAC 2PC in this order... .)
For me it says:
RGL mesh source: RegularMeshesAndSkinnedMeshes
[Thu Jul 11 14:47:59 2024]: Logging configured: level=info, file=(disabled), stdout=true
[14:47:59][ 13 us][info]: RGL Version 0.16.2 branch=develop commitSHA1=8eee9475ced107603063533e6333498252dd0ba8
[14:47:59][ 18572 us][info]: Running on GPU: NVIDIA GeForce RTX 3080
[14:47:59][ 23 us][info]: Built against OptiX SDK version: 7.2.0
[14:47:59][ 4 us][info]: Built against OptiX ABI version: 41
[14:47:59][ 4 us][info]: Built against CUDA Toolkit version: 11.7
[14:47:59][ 9 us][info]: Installed CUDA runtime version: 11.7
[14:47:59][ 4 us][info]: Installed CUDA driver version: 12.3
[14:47:59][ 89019 us][info]: Installed NVidia kernel driver version: 545.29.06
Full /home/mfc/.config/unity3d/AWF/AWSIM Labs/Player.log
: https://gist.github.com/xmfcx/a1e70ac4e0c00f980f5c207a64726ab2
AWSIM Labs or AWSIM probably won't run on WSL2. @lchojnack @WJaworskiRobotec @msz-rai are from Robotec and their team developed the Lidar Unity simulator. Maybe they can help but seeing awsim also slow, I wouldn't recommend it.
@xmfcx By looking on the errors, my quick intuition is that either CUDA driver support is not full on WSL2, or implicit initialization behavior of the CUDA driver is different on WSL2 than everywhere else and it doesn't get automatically initialized before constructing OptiX context. If you're willing to recompile RGL and debug, the relevant place to look is here.
Maybe they can help but seeing awsim also slow, I wouldn't recommend it.
Would you mind expanding your thought?
Thanks for the gist to compare. I have to solve it so i will do an deepdive.
Was quite a journey to get it to this point.
1.install WSL2 with Ubuntu
Install WSLg
set up Hardware-accelerated vulkan in WSL2 · sudo add-apt-repository ppa:kisak/kisak-mesa sudo apt update sudo apt upgrade
AWSIM Labs Quickstart setup
to get Autoware WSL - AWSIM WSL 1PC
There are some test on 4 pages from last year.... https://www.phoronix.com/review/windows11-wsl2-zen4/2
the WSL2 vs native performance doesn t look so bad.
There was also an CUDA GPU article. I post it when i refound it.
@sasa42 could you also share the
nvidia-smi
results with us?don’t install the nvidia driver, there’s no real nvidia gpu available in WSL2.
my output:
mfc@mfc-leo:~$ nvidia-smi Thu Jul 11 13:32:27 2024 +---------------------------------------------------------------------------------------+ | NVIDIA-SMI 545.29.06 Driver Version: 545.29.06 CUDA Version: 12.3 | |-----------------------------------------+----------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+======================+======================| | 0 NVIDIA GeForce RTX 3080 Off | 00000000:2D:00.0 On | N/A | | 53% 75C P2 148W / 320W | 3447MiB / 10240MiB | 23% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ +---------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=======================================================================================| | 0 N/A N/A 1583 G /usr/lib/xorg/Xorg 838MiB | | 0 N/A N/A 5867 G ...yOnDemand --variations-seed-version 42MiB | | 0 N/A N/A 6345 G ...,262144 --variations-seed-version=1 226MiB | | 0 N/A N/A 1131802 G ...erProcess --variations-seed-version 23MiB | | 0 N/A N/A 1164733 G ...erProcess --variations-seed-version 65MiB | | 0 N/A N/A 2589250 C+G ./awsim_labs_v1.1.1/awsim_labs.x86_64 2198MiB | | 0 N/A N/A 3169233 G ...seed-version=20240708-050118.992000 15MiB | +---------------------------------------------------------------------------------------+
On the left i have some detailed debug output...
(https://github.com/autowarefoundation/AWSIM-Labs/assets/1211326/1b06fdc4-380a-4861-8048-593c5baab9c2)
Do you need the nvidia-smi, too
@prybicki
Maybe they can help but seeing awsim also slow, I wouldn't recommend it.
Would you mind expanding your thought?
He has a pc with high end parts: 7945HX cpu and rtx 4090 yet he has very low framerates. I've assumed this was due to running in WSL2, and since it is running slow, maybe it will keep running slow even if lidar simulation is fixed. So this is why I don't recommend running AWSIM in WSL2.
Maybe we should fix the problem of AWSIM Labs not running on windows.
Then he could run autoware on wsl2 and awsim labs on native windows.
nvidia-smi...
Background
Problem
All CUDA, TensorRT, cuDNN versions are correct, dds settings are done, correct version of autoware is compiled etc.
But on AWSIM Labs side, there is no lidar simulation, no red point cloud visualization, lidar topics are empty.
Report
... some lidar are missing.... and the pointcloud_raw_ex
Originally posted by @sasa42 in https://github.com/autowarefoundation/AWSIM-Labs/issues/87#issuecomment-2222514692