GPU 0 does not support virtual memory - (core dumped)

[hg1]

Hi There,

I get this error “GPU 0 does not support virtual memory” on Orin AGX,
yet Orin AGX, with 2048 Cuda cores, supports virtual memory, right?

It appears that the error is stem from tiny-cuda-nn library (output below)
I tested and got the same error on the NVlabs/tiny-cuda-nn repo

tiny-cuda-nn warning: GPUMemoryArena: GPU 0 does not support virtual memory. Falling back to regular allocations, which will be larger and can cause occasional stutter.
root@a520461b9c4d:/main/tiny-cuda-nn# ./build/mlp_learning_an_image data/images/albert.jpg data/config_hash.json
Loading custom json config ‘data/config_hash.json’.
Beginning optimization with 10000000 training steps.
tiny-cuda-nn warning: GPUMemoryArena: GPU 0 does not support virtual memory. Falling back to regular allocations, which will be larger and can cause occasional stutter.
terminate called after throwing an instance of ‘std::runtime_error’
what(): /main/tiny-cuda-nn/include/tiny-cuda-nn/cuda_graph.h:99 cudaStreamEndCapture(stream, &m_graph) failed: operation failed due to a previous error during capture
Aborted (core dumped)

btw, I’m using Cuda -12.2
Built on Tue_Jun_13_19:22:54_PDT_2023
Cuda compilation tools, release 12.2, V12.2.91
Build cuda_12.2.r12.2/compiler.32965470_0

Any idea how to solve this?

[AastaLLL]

Hi,

It’s a warning message and might not be the reason causes the failure.

The GitHub is verified on a desktop GPU.
Based on the issue opened below, it seems not supported Jetson officially.

Issues · NVlabs/tiny-cuda-nn

Thanks.