GPU crash on Hello World example

My system:

  • CPU: x86_64
  • GPU: RTX 4090
  • OS: Ubuntu 20.04
  • Nvidia Driver: 535.104.05 (installed via .run file)

Problem:
First of all, latest containerized Isaac Sim (built via IsaacSim-dockerfiles) has no curl installed, which should probably be fixed, because user can’t run Isaac Sim with --rm flag anymore (you have to install curl and restart the same container).
Then Isaac Sim starts without errors, just a lot of warnings, but ALL Isaac Examples (even “Hello World”) crashes the program with such error:

[Error] [carb.graphics-vulkan.plugin] GPU crash is detected. Crash dump is written into: /root/.nvidia-omniverse/logs/Kit/Isaac-Sim/2022.2/kit_20230908_091716.nv-gpudmp
[Error] [carb.graphics-vulkan.plugin] GPU crash dump is successfully written
[Error] [carb.graphics-vulkan.plugin] VkResult: ERROR_DEVICE_LOST
[Error] [carb.graphics-vulkan.plugin] submitToQueueCommon failed.
[Error] [gpu.foundation.plugin] A GPU crash occurred. Exiting the application...
Reasons for the failure: a device lost, out of memory, or an unexpected bug.
[Error] [carb.crashreporter-breakpad.plugin] crash detected
... a bunch of warnings here ...
[Fatal] [carb.crashreporter-breakpad.plugin] Thread 251 backtrace follows:
[Fatal] [carb.crashreporter-breakpad.plugin] 000: libpthread.so.0!funlockfile+0x60
[Fatal] [carb.crashreporter-breakpad.plugin] 001: libgpu.foundation.plugin.so!carbGetPluginDeps+0x2b3f
[Fatal] [carb.crashreporter-breakpad.plugin] 002: libgpu.foundation.plugin.so!void std::vector<std::tuple<char const*, float, char const*, unsigned int>, std::allocator<std::tuple<char const*, float, char const*, unsigned int> > >::_M_realloc_insert<std::tuple<char const*, float, char const*, unsigned int> >(__gnu_cxx::__normal_iterator<std::tuple<char const*, float, char const*, unsigned int>*, std::vector<std::tuple<char const*, float, char const*, unsigned int>, std::allocator<std::tuple<char const*, float, char const*, unsigned int> > > >, std::tuple<char const*, float, char const*, unsigned int>&&)+0x1168f
[Fatal] [carb.crashreporter-breakpad.plugin] 003: libgpu.foundation.plugin.so!void std::vector<std::tuple<char const*, float, char const*, unsigned int>, std::allocator<std::tuple<char const*, float, char const*, unsigned int> > >::_M_realloc_insert<std::tuple<char const*, float, char const*, unsigned int> >(__gnu_cxx::__normal_iterator<std::tuple<char const*, float, char const*, unsigned int>*, std::vector<std::tuple<char const*, float, char const*, unsigned int>, std::allocator<std::tuple<char const*, float, char const*, unsigned int> > > >, std::tuple<char const*, float, char const*, unsigned int>&&)+0x1b9b9
[Fatal] [carb.crashreporter-breakpad.plugin] 004: libomni.kit.renderer.plugin.so!carbOnPluginPostShutdown+0x2af5
[Fatal] [carb.crashreporter-breakpad.plugin] 005: libomni.kit.renderer.plugin.so!carbOnPluginPostShutdown+0x2ca9
[Fatal] [carb.crashreporter-breakpad.plugin] 006: libomni.kit.renderer.plugin.so!std::__future_base::_State_baseV2::_M_is_deferred_future() const+0x8f
[Fatal] [carb.crashreporter-breakpad.plugin] 007: libcarb.events.plugin.so!std::_Hashtable<unsigned long, unsigned long, std::allocator<unsigned long>, std::__detail::_Identity, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, true, true> >::_M_insert_unique_node(unsigned long, unsigned long, std::__detail::_Hash_node<unsigned long, false>*)+0x188c
[Fatal] [carb.crashreporter-breakpad.plugin] 008: libcarb.events.plugin.so!std::_Hashtable<unsigned long, unsigned long, std::allocator<unsigned long>, std::__detail::_Identity, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, true, true> >::_M_insert_unique_node(unsigned long, unsigned long, std::__detail::_Hash_node<unsigned long, false>*)+0x2026
[Fatal] [carb.crashreporter-breakpad.plugin] 009: libomni.kit.loop-default.plugin.so!std::thread::_State_impl<std::thread::_Invoker<std::tuple<omni::kit::RunLoopThread::run()::{lambda()#1}> > >::~_State_impl()+0x8ac
[Fatal] [carb.crashreporter-breakpad.plugin] 010: libomni.kit.loop-default.plugin.so!std::thread::_State_impl<std::thread::_Invoker<std::tuple<omni::kit::RunLoopThread::run()::{lambda()#1}> > >::~_State_impl()+0xe8e3
[Fatal] [carb.crashreporter-breakpad.plugin] 011: libomni.kit.loop-default.plugin.so!std::thread::_State_impl<std::thread::_Invoker<std::tuple<omni::kit::RunLoopThread::run()::{lambda()#1}> > >::_M_run()+0x1d
[Fatal] [carb.crashreporter-breakpad.plugin] 012: libstdc++.so.6!std::error_code::default_error_condition() const+0x34
[Fatal] [carb.crashreporter-breakpad.plugin] 013: libpthread.so.0!start_thread+0xd9
[Fatal] [carb.crashreporter-breakpad.plugin] 014: libc.so.6!clone+0x43
./runapp.sh: line 4:    14 Segmentation fault      (core dumped) /isaac-sim/isaac-sim.sh --/persistent/isaac/asset_root/default="$OMNI_SERVER" --allow-root $@

My goal:
I try to figure out how to use Isaac ROS, so my first step was to build it and run benchmarks (done), now my second step: run ANY Isaac ROS example with Isaac Sim.

P.S. Isaac Sim itself works, even Isaac Assets work fine in it (I could spawn various robots without any error).

@divelix i am just another user passing by. is it possible for you to upload the most recent log in its entirety? it should provide more context to the mods/devs when they get to this thread. you should be able to find it in the Issac Sim subfolder in ~/.nvidia-omniverse/logs/Kit/

does the Hello World example crash when you hit the “Load” button in the World Controls rollout or after a specific action?

Thank you for reply! Crash happens when I click LOAD button of Hello World example.

Log file:
kit_20230919_172308.log (1.3 MB)

P.S. This error happens with both containerized and Workstation installed Isaac Sims on my machine. Log is for workstation one.