search

mlco2 / codecarbon

Track emissions from Compute and recommend ways to reduce their impact on the environment.
https://mlco2.github.io/codecarbon
MIT License
1k stars 157 forks source link

Nividia TX2 support ? #525

Open qxpBlog opened 2 months ago

qxpBlog commented 2 months ago

@dsblank @Lothiraldan @ndevenish @mp15 @guylifshitz My environment is NVIDIA TX2， when use the package codecarbon to get information of GPU，but it can not find GPU:

[codecarbon INFO @ 21:03:55] [setup] RAM Tracking...
[codecarbon INFO @ 21:03:55] [setup] GPU Tracking...
[codecarbon INFO @ 21:03:55] No GPU found.
[codecarbon INFO @ 21:03:55] [setup] CPU Tracking...
[codecarbon WARNING @ 21:03:55] No CPU tracking mode found. Falling back on CPU constant mode.
huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...
To disable this warning, you can either:
    - Avoid using `tokenizers` before the fork if possible
    - Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)
[codecarbon WARNING @ 21:03:55] We saw that you have a ARMv8 Processor rev 1 (v8l) but we don't know it. Please contact us.
[codecarbon INFO @ 21:03:55] CPU Model on constant consumption mode: ARMv8 Processor rev 1 (v8l)
[codecarbon INFO @ 21:03:55] >>> Tracker's metadata:
[codecarbon INFO @ 21:03:55]   Platform system: Linux-5.10.104-tegra-aarch64-with-glibc2.17
[codecarbon INFO @ 21:03:55]   Python version: 3.8.13
[codecarbon INFO @ 21:03:55]   CodeCarbon version: 2.3.4
[codecarbon INFO @ 21:03:55]   Available RAM : 6.329 GB
[codecarbon INFO @ 21:03:55]   CPU count: 6
[codecarbon INFO @ 21:03:55]   CPU model: ARMv8 Processor rev 1 (v8l)
[codecarbon INFO @ 21:03:55]   GPU count: None
[codecarbon INFO @ 21:03:55]   GPU model: None

But the result of torch.cuda.is_available() is true， so i want to know if ` codecarbon could suport the facility TX2? Looking forward to your reply.

benoit-cty commented 2 months ago

Thanks for reporting.

TX2 is a quite specific so maybe it lack support from pynvml.

Can you try with debug mode ?

export CODECARBON_LOG_LEVEL=DEBUG
codecarbon monitor

So we may see more info in the log that prevent CodeCarbon from reading GPU info.

benoit-cty commented 2 weeks ago

Hello, can you try with https://github.com/mlco2/codecarbon/pull/571 ?