NVIDIA Audio SDK

muthu3 · July 17, 2021, 1:52pm

Hi,

I cannot get to work the sample codes in the audio SDK. It always shows up with an error on NvAFX_Run() failed writes out empty audio files instead of filtered audio files. Has anyone faced such a situation ?

Regards,
Muthu

ppang · July 19, 2021, 10:32pm

Are referring to the Audio Effects SDK?
https://developer.nvidia.com/maxine-getting-started

muthu3 · July 20, 2021, 4:25am

Yes

lingq · July 20, 2021, 6:54pm

Can you share which version, what GPU do you run on?

muthu3 · July 21, 2021, 2:34am

HI @lingq,I am trying to run on GCP with Tesla T4 GPUs. Thanks

lingq · July 21, 2021, 3:35pm

It might be dependency and driver issues. We have a containerized version that could make it easier for your deployment. Could you please give this a try and let me know how it goes for you? Maxine Audio Effects SDK version 1.0 | NVIDIA NGC

Thanks,
Alex

muthu3 · July 22, 2021, 6:41am

Hi @lingq !

Thanks for the suggestions. I cleared all the authorization protocols and managed to obtain the image. I am trying to run the image with the following command:

docker run --gpus all -it --rm -v /home/jupyter:/usr/local/AudioFX nvcr.io/nvidia/maxine/maxine-audioeffects-sdk:5.5

But what turns out was this:
docker: Error response from daemon: OCI runtime create failed: container_linux.go:367: starting container process caused: process_linux.go:495: container init caused: Running hook #0:: error running hook: exit status 1, stdout: , stderr: nvidia-container-cli: initialization error: nvml error: driver/library version mismatch: unknown.

The following is the driver specs:
lsmod | grep nvidia

nvidia_uvm 987136 0
nvidia_drm 53248 3
nvidia_modeset 1228800 3 nvidia_drm
nvidia 34131968 271 nvidia_uvm,nvidia_modeset
drm_kms_helper 188416 1 nvidia_drm
drm 491520 6 drm_kms_helper,nvidia_drm

I think it is a driver or a compute capability issue and we are using a Tesla T4. It would be nice if you suggest a step forward.

Many Thanks,
Muthu

lingq · July 22, 2021, 6:59am

@Tanay Varshney could you help take a pass on it?

Thanks!
Alex

tvarshney · July 22, 2021, 3:51pm

Hi @muthu3,

Could you share the docker and nvidia docker versions you have installed?

muthu3 · July 23, 2021, 9:35am

@tvarshney Thanks for stepping in. I have got

Docker version 20.10.7, build f0df350

for both docker and nvidia-docker

tvarshney · July 26, 2021, 3:21pm

@muthu3

Can you please share the output of:

“nvidia-smi”
and
“docker run --gpus=all --rm nvcr.io/nvidia/cuda:11.1-base nvidia-smi”

muthu3 · July 26, 2021, 4:35pm

±----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
±----------------------------------------------------------------------------+

and

±----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
±----------------------------------------------------------------------------+

respectively

tvarshney · July 26, 2021, 5:01pm

Could you re-provision the machine and install cuda from here?: https://developer.nvidia.com/cuda-11.1.1-download-archive?target_os=Linux&target_arch=x86_64&target_distro=Ubuntu&target_version=1804&target_type=deblocal

lingq · July 28, 2021, 5:53pm

@Tanay Varshney were you able to look into this yet?

muthu3 · July 28, 2021, 6:18pm

@lingq @tvarshney Thanks a lot guys! I am now able to host the image and run some sample codes on my Tesla T4 instance successfully

lingq · July 28, 2021, 7:42pm

Thank you Muthu! Can you share more feedback on how we can make it easier for developer like you so it’s easier to setup and get it running? looks like it’s the cuda version mistmatch?

Btw - what’s the use case you are building?

muthu3 · August 5, 2021, 3:26am

Hi @lingq , Sorry for the late response. Yes, I guess it was due to the CUDA version mismatch. However I found recently that there are easy options to switch the version of our liking in a GCP instance. I should have read through the docs carefully.

We are building a VR chat room with customized capabilities and data-driven cognitive experience. Thanks a lot for your assistance

Regards,
Muthu

lingq · August 5, 2021, 4:30am

Thank you Muthu. Let us know how it goes for you.

Do you mind sharing which company you are working for?

system · October 4, 2021, 4:31am

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.