TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
MindStudio’s tests on an MSI Prestige RTX Spark laptop found native Linux generated 13.9 tokens per second on GPT-OSS-120B, compared with 65.4 on Windows and 65.1 in WSL. The report links the gap to low GPU clocks under Linux and power-management firmware that Linux cannot currently read; Nvidia has not provided a response in the supplied material.
Native Linux inference on an MSI Prestige RTX Spark laptop produced 13.9 tokens per second in MindStudio’s GPT-OSS-120B test, far below the 65.4 tokens per second on Windows and 65.1 in Windows Subsystem for Linux (WSL). The report attributes the slowdown to Linux running the GPU at a much lower clock speed, linked to power-management firmware that the Linux driver cannot currently access.
MindStudio reported that the laptop’s GPU ran at about 790 MHz during its Linux benchmark, compared with roughly 2.2 GHz under Windows. In a longer run, the report said Windows completed the workload in about eight minutes, while native Linux took 37 minutes. Temperatures were between 40 and 47 degrees Celsius, which MindStudio cited as evidence that heat was not causing the reduced speed.
The tests also compared the laptop with Nvidia’s DGX Spark desktop system, which reached 61 tokens per second on the same model, according to the report. In a separate vLLM test with eight concurrent clients, MindStudio recorded 127 tokens per second in WSL and 39.5 on native Linux. These results come from the report’s test setup and are not a broad, independently verified benchmark of every RTX Spark laptop or Linux configuration.
Linux did run on the tested machine, but setup required workarounds. MindStudio described installing Linux, enabling the open Nvidia kernel driver and changing a memory allocation setting in Windows. The laptop initially exposed only 28GB of its 128GB unified memory to Linux; after changing the Windows-side GPU reservation, Linux could see 120GB, the report said.
Linux Performance Falls Behind Windows
The measured gap matters to developers considering these laptops for local AI workloads. Many AI tools, including vLLM and SGLang, are designed primarily for Linux environments, while the RTX Spark laptops are sold with Windows on ARM. If native Linux cannot use the GPU at its expected clock speed, users may have to choose between a preferred software environment and the performance they paid for.
MindStudio’s results suggest WSL is a practical workaround for this specific test: it ran Linux tools while retaining performance close to Windows. That may suit some developers, but it is not the same as running Linux directly on the hardware. The findings also point to a platform-support issue rather than an inherent limitation of the GPU, though the report’s explanation has not been independently confirmed by Nvidia in the supplied material.
Products Worth Considering
Apple MacBook Air 13.6-inch with M5 Chip, 16GB Memory, 512GB SSD, Sky Blue
Apple MacBook Air 15-inch with M5 chip, 24GB RAM, 512GB SSD, Midnight
Amazon Kindle Paperwhite (2024), 16 GB, 7-inch Display, Black
As an affiliate, we earn on qualifying purchases.
How the Laptop Runs Linux
The RTX Spark laptop line includes systems from manufacturers such as MSI, Asus and Dell. The supplied report says these machines use the same GPU design as Nvidia’s DGX Spark desktop, but ship with Windows on ARM; the desktop system runs Nvidia’s Linux build. The laptop’s unified memory, which is shared between CPU and GPU and can be configured up to 128GB, is part of its appeal for running large AI models locally.
MindStudio said its Linux installation needed hardware-specific support and configuration. A Debian installation was more reliable than an initial attempt with Omarchy, but the installer lacked wireless networking and did not initially recognize several laptop components. The report said an Omarchy kernel with Nvidia’s open-source driver eventually enabled a desktop, CUDA acceleration and 4K output at 120Hz. The result shows that Linux can operate on the tested laptop, but not that all models or distributions will work identically.
According to MindStudio, the power-management issue involves Nvidia’s Platform Controllers and Framework, a Windows-only component that manages power delivery between the CPU and GPU. The report says the relevant power table is encrypted and can be decrypted by the Windows driver, while Linux lacks an equivalent implementation. This is the report’s explanation for the clock-speed difference, not a technical statement independently confirmed here by Nvidia.
“Windows produced 65.4 tokens per second. WSL … hit 65.1 tokens per second. … Native Linux on the laptop … managed only 13.9.”
— MindStudio, in its report
Firmware Access Remains Unresolved
The source material does not include a response from Nvidia, confirmation of the firmware explanation by the company, or a commitment to add equivalent Linux power-management support. It also does not provide the report’s publication date, full benchmark methodology, or enough information to assess how results might vary across Linux distributions, driver versions, laptop makers or workloads.
The supplied text begins to describe a different performance pattern when the laptop is unplugged but cuts off before giving the result. It is therefore unclear from the material provided how battery operation changes the Linux-versus-Windows comparison. The reported benchmark figures should be treated as measurements from one described test setup, not universal performance guarantees.
Nvidia Support Is the Key Milestone
MindStudio said a lasting fix would require Nvidia to bring the platform controller’s power-management logic to its Linux driver. No such fix, release date or public Nvidia commitment appears in the supplied source. Until that changes, the report identifies WSL as the available performance-preserving option for users who want to run Linux-based AI tools on the tested laptop.
Further testing across more RTX Spark laptop models, Linux distributions and driver releases could show whether the same limitation is widespread. For now, buyers should distinguish between Linux being installable and Linux delivering the same GPU performance as Windows: MindStudio’s tests found the former, but not the latter.
Key Questions
How fast was native Linux in MindStudio’s test?
MindStudio measured 13.9 tokens per second running GPT-OSS-120B on native Linux on the tested MSI Prestige RTX Spark laptop.
How did Windows and WSL compare?
The report recorded 65.4 tokens per second on Windows and 65.1 in WSL in the same model test. Those are results from MindStudio’s setup, not a guarantee for all workloads.
Why does the report say Linux was slower?
MindStudio linked the result to GPU clocks of about 790 MHz on Linux, compared with roughly 2.2 GHz on Windows, and attributed the difference to power-management firmware unavailable to the Linux driver. Nvidia’s confirmation is not included in the supplied material.
Can Linux run on an RTX Spark laptop?
Yes, according to MindStudio’s account of its MSI Prestige test, though installation and hardware configuration required workarounds. Its report does not establish that every laptop model or distribution will behave the same way.
Is there a confirmed fix?
No confirmed fix or release schedule is provided. MindStudio said resolving the issue would require Nvidia to add the platform controller’s power-management logic to its Linux driver.
Source: mindstudio.ai
