Skip to content
MindStudio·

🐧Linux Runs Nvidia RTX Spark Laptops But GPU Crawls at 780 MHz

Your Linux laptop GPU is running at a third of its speed

TL;DR

Linux runs fine on Nvidia RTX Spark laptops, but the GPU idles around 780 MHz versus 2.2 GHz on Windows. Native Linux manages 13.9 tokens per second on GPT-OSS-120B, a benchmark Windows finishes in 8 minutes that Linux needs 37 minutes to complete.

Linux runs on Nvidia's RTX Spark laptops, and the GPU works through the open kernel driver. The catch: performance falls off a cliff. The GPU clock sits around 780 MHz on Linux when plugged in, while Windows hits about 2.2 GHz. If you're buying this laptop for local LLM work on Linux, that gap matters. Native Linux produces 13.9 tokens per second with GPT-OSS-120B, against 65.4 on Windows. WSL is the workaround, since it runs on the Windows driver stack and keeps full GPU boost even on battery. The root cause is the Nvidia Platform Controllers and Framework, which manages power delivery between CPU and GPU. Its power table is encrypted, and only the Windows driver can decrypt it. Native Linux also sees only 28GB of the 128GB unified memory, because the firmware reserves 96GB for the GPU and only Windows exposes the setting to change that split. A real fix means Nvidia has to port the platform controller's power logic into its Linux driver. Until that happens, WSL is the practical path.

Linux Runs Nvidia RTX Spark Laptops But GPU Crawls at 780 MHz — MindStudio

Key Points

1

Native Linux delivers 13.9 tokens per second on GPT-OSS-120B, while Windows hits 65.4 tokens per second on the same model and hardware setup.

Keep reading — it's free

Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.

Also get
or

Already a member? Sign in

nvidialinuxwslgpu-accelerationllm-inference

Comments

Subscribe to join the conversation...

Be the first to comment