Nvidia RTX Spark Laptops Run Linux at 35% Speed Without Windows
Nvidia's RTX Spark laptops, including the MSI Prestige N16 Flip AI Plus, can run Linux, but GPU performance is severely limited: plugged in, the GPU locks to about 780 MHz instead of 2.2 GHz, dropping AI inference from 65 to 14 tokens per second. The cause is an encrypted power-management firmware that only the Windows driver can read.
The 35% problem: Linux runs, but slowly
Linux installs and runs on RTX Spark laptops like the MSI Prestige N16 Flip AI Plus, complete with GPU acceleration via Nvidia's open kernel driver. The catch is clock speed: once the laptop is plugged into wall power, the GPU gets locked to roughly 780 MHz on Linux instead of the 2 GHz-plus it hits on Windows.
That turns a chip capable of around 65 tokens per second on a 120 billion parameter model into one struggling to break 14 tokens per second, for reasons tied to power management firmware that Linux currently can't read.
The culprit is a hidden Windows-only component called the Nvidia Platform Controllers and Framework, which manages power delivery between CPU and GPU. It has no Linux equivalent, and its power table is encrypted so only the Windows driver can decrypt it.
Benchmarks on GPT-OSS-120B make the gap plain: Windows hit 65.4 tokens per second, WSL hit 65.1, the Nvidia DGX Spark desktop hit 61, and native Linux on the same laptop managed only 13.9. With vLLM running eight concurrent clients, WSL reached 127 tokens per second while native Linux reached 39.5.
Nvidia RTX Spark laptop: key specs and performance
| Spec | Detail |
|---|---|
| GPU clock on Windows (plugged in) | ~2.2 GHz |
| GPU clock on Linux (plugged in) | ~780 MHz |
| GPU clock on Linux (battery) | ~1,800 MHz |
| AI inference (Windows, GPT-OSS-120B) | 65.4 tokens/sec |
| AI inference (WSL, GPT-OSS-120B) | 65.1 tokens/sec |
| AI inference (native Linux, GPT-OSS-120B) | 13.9 tokens/sec |
| Unified memory (max) | 128 GB |
| Default GPU memory reservation | 96 GB |
Why the throttling happens (and why it's not thermal)
The slowdown isn't a cooling issue. Temperatures stayed between 40 and 47°C during the Linux benchmark run, so the GPU simply isn't being allowed to run at full speed.
Oddly, the direction of the problem flips depending on power source. Unplugged, Linux hit around 61 to 67 tokens per second on LM Studio, comparable to Windows. Plugged in, that same test dropped to about 15 tokens per second. Clock speed told the same story: around 1,800 MHz unplugged under load, and down to roughly 780 MHz when connected to wall power.
Memory allocation is a separate headache. The laptop reserves 96GB of unified memory for the GPU by default, and only Windows exposes a setting to change that split. Before Linux can see more than a sliver of RAM, you have to boot into Windows and drop the GPU reservation down to 4GB, which frees up 120GB for Linux.
Getting Linux installed at all is not plug-and-play. A first attempt using Omarchy (an Arch-based distribution) failed on the MSI Prestige with a flickering, unstable display during boot. Switching to Debian worked more reliably but still needed workarounds: the installer couldn't connect to Wi-Fi, so installation required a USB Ethernet adapter, and vanilla Debian didn't recognize the keyboard, touchpad, or GPU. Going back to Omarchy solved the driver problem because its kernel already bundled Nvidia's open-source driver.
Nvidia's decision to encrypt the power table and keep the platform controller Windows-only is a bad look for a platform marketed at AI developers. The company should either port the logic to Linux or document the limitation clearly. Until then, WSL is the workaround, but it's not a substitute for native Linux support.
What actually works: WSL is the practical middle ground
For now, Windows Subsystem for Linux (WSL) is the best way to run Linux-first AI tools on these machines without losing performance.
WSL rides on the Windows driver stack and keeps full GPU boost whether the laptop is plugged in or on battery, while still running tools like vLLM. That makes it the pragmatic choice for developers who want Linux tooling but need the GPU at full speed.
A real fix depends on Nvidia, specifically on porting the platform controller's power management logic into its Linux driver. That would effectively turn the laptop into a portable version of the DGX Spark, which ships with Nvidia's own Linux build out of the box and hits 61 tokens per second.
The RTX Spark platform, which also appears in machines like the Asus ProArt and Dell XPS, uses the same GPU design as the DGX Spark desktop unit. But where the DGX Spark runs Linux natively, every laptop variant ships with Windows on ARM. That distinction matters because Windows on ARM with CUDA support is new, and the laptops support unified memory configurations up to 128GB shared between CPU and GPU.
What this means for AI developers
Developers who buy an RTX Spark laptop hoping to ditch Windows for a Linux-first AI workflow are in for a disappointment—at least until Nvidia addresses the driver gap.
The performance penalty is steep: native Linux runs at roughly 35% of Windows speed when plugged in, and the gap widens under concurrent workloads. For anyone running large models locally, that's the difference between a usable development machine and a paperweight.
The encryption of the power table also raises transparency concerns. Developers can't inspect or modify the firmware to work around the issue, which leaves them dependent on Nvidia to ship a fix. Until then, WSL remains the only way to get both Linux tooling and full GPU performance on these laptops.
Availability and pricing
RTX Spark laptops are available now from manufacturers including MSI, Asus, and Dell. Pricing varies by configuration and region; check with local retailers for exact pricing in the US, UK, and EU.
The Linux throttling issue affects all RTX Spark laptop models, not just the MSI Prestige. Nvidia has not announced a timeline for a fix.
The RTX Spark laptops are powerful AI machines on Windows, but Linux users are second-class citizens for now. Until Nvidia opens up its power-management firmware or ports the platform controller to Linux, developers who need native Linux performance should look elsewhere—or stick with WSL.
Frequently asked questions
Can I run Linux on an Nvidia RTX Spark laptop?
Yes, Linux installs and runs with GPU acceleration via Nvidia's open kernel driver, but performance is severely limited when plugged in due to encrypted power-management firmware.
Why is Linux so slow on RTX Spark laptops?
The GPU clock locks to about 780 MHz on Linux when plugged in, versus 2.2 GHz on Windows, because a Windows-only component called the Nvidia Platform Controllers and Framework manages power delivery and its firmware is encrypted.
What is the best way to run Linux AI tools on an RTX Spark laptop?
Use Windows Subsystem for Linux (WSL). It delivers near-native Windows performance (65.1 vs 65.4 tokens/sec) and keeps full GPU boost whether plugged in or on battery.
Sources
- Nvidia Open Kernel Driver. Nvidia, 2026. https://github.com/NVIDIA/open-gpu-kernel-modules
- Nvidia DGX Spark. Nvidia, 2026. https://www.nvidia.com/en-us/products/workstations/dgx-spark/
- Linux on the Nvidia RTX Spark Laptop: What Actually Works | MindStudio. news.ycombinator.com, 2026. https://www.mindstudio.ai/blog/linux-nvidia-rtx-spark-laptop