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 reported that native Linux on an MSI Prestige RTX Spark laptop generated 13.9 tokens per second on GPT-OSS-120B, compared with 65.4 on Windows and 65.1 in WSL. The report attributes the gap to Linux lacking access to a Windows-only Nvidia power-management component; Nvidia has not provided a response in the supplied material.
Native Linux inference on an MSI Prestige RTX Spark laptop ran at 13.9 tokens per second in a MindStudio test, compared with 65.4 on Windows, with the report linking the gap to Linux’s inability to access Nvidia’s Windows-only power-management component. The result suggests that developers who install Linux directly on the laptop may lose much of its AI performance, even though GPU acceleration works.
MindStudio tested the laptop with GPT-OSS-120B. It recorded 65.4 tokens per second on Windows and 65.1 in Windows Subsystem for Linux (WSL), while the same laptop running native Linux reached 13.9. The report also lists 61 tokens per second for Nvidia’s DGX Spark desktop, though that is a separate system and not a like-for-like laptop configuration.
The report says GPU clocks on Linux while plugged into wall power stayed near 780 MHz, compared with roughly 2.2 GHz on Windows. In a longer run, it says Windows completed the workload in about eight minutes and Linux took 37 minutes. Temperatures were reported at 40–47°C, leading the author to attribute the slowdown to power limits rather than thermal throttling.
A second test using vLLM and eight concurrent clients also favored WSL: MindStudio reported 127 tokens per second there and 39.5 on native Linux. The article says Linux could run with Nvidia’s open kernel driver and support a desktop, CUDA acceleration and 4K output at 120Hz after setup. Those capabilities did not prevent the lower performance reported in the plugged-in benchmark.
Linux Users Face a Large Inference Penalty
The gap matters to developers considering these laptops for locally run AI models. A machine that delivers around 65 tokens per second under Windows or WSL but 13.9 under native Linux may be substantially less useful for interactive model work when booted into Linux, based on MindStudio’s test.
The results also distinguish software compatibility from performance parity. Linux booted and used the GPU, but the report says its power-management path did not let the GPU reach the clocks observed on Windows. For users who depend on Linux-first tools, WSL may offer a compromise: Linux applications run in a Linux environment while using Windows’ driver stack. The findings come from one reported laptop and workload, so they should not be treated as a measure of every RTX Spark configuration or AI task.
Products Worth Considering
Apple MacBook Air 13-inch with M5 chip, 16GB memory, 512GB SSD, and 12MP Center Stage camera
Samsung Galaxy Tab A11+ 11-Inch Tablet, 6GB RAM, 128GB Storage, Expandable, Gray
Apple MacBook Air 15-inch with M5 chip, 16GB Memory, 512GB SSD, and Liquid Retina Display
As an affiliate, we earn on qualifying purchases.
Why the RTX Spark Runs Linux Differently
The RTX Spark laptop platform is described in the report as pairing an Nvidia GPU with unified memory shared by the CPU and GPU, in configurations up to 128GB. Models cited include the MSI Prestige, Asus ProArt and Dell XPS. The laptops ship with Windows on ARM, while the DGX Spark desktop comes with Nvidia’s Linux build, making the laptop’s operating system choice relevant to developers who use Linux-based AI software.
MindStudio says it needed several workarounds to install and configure Linux on the MSI. A Debian installation lacked some laptop hardware support, while the author later used a distribution with Nvidia’s open kernel driver. The laptop initially exposed only 28GB of its 128GB unified memory to Linux because firmware reserved 96GB for the GPU. The report says the memory split had to be changed in Windows before Linux could see more memory.
According to the article, the power issue involves Nvidia Platform Controllers and Framework, a Windows-only component that manages power delivery between the CPU and GPU. MindStudio says the relevant firmware power table is encrypted and that only the Windows driver can decrypt it. The supplied source does not include a statement from Nvidia confirming that explanation.
“Windows produced 65.4 tokens per second. WSL hit 65.1 tokens per second. Native Linux on the laptop managed only 13.9.”
— MindStudio report
Nvidia’s Linux Support Remains Unconfirmed
The supplied report does not include a response from Nvidia, confirmation of when or whether the company will add equivalent power-management support to Linux, or a broader set of tests across laptop models. The benchmark results are attributed to MindStudio and cover specified workloads; performance may differ with other models, drivers, software versions and power settings.
The source’s excerpt ends while discussing how performance changes when the laptop is unplugged, so the full battery-powered results are not available here. It is also unclear whether the encrypted firmware behavior applies in the same way to every RTX Spark laptop. The explanation linking that firmware to the observed clock limits is the report’s account, not an independently verified finding in the supplied material.
Linux Performance Depends on Nvidia
For now, MindStudio presents WSL as the practical workaround for users who want Linux tools without giving up the Windows driver’s reported GPU boost. A lasting native-Linux fix, according to the report, would require Nvidia to provide Linux with the relevant platform power-management logic.
No release, driver update or Nvidia timetable is identified in the supplied source. Readers should watch for an Nvidia response or Linux driver changes, and for follow-up testing on other RTX Spark laptops and workloads. Until then, the reported results describe a substantial limitation in one native-Linux setup, not a confirmed performance rule for the entire product line.
Key Questions
How fast was native Linux in the reported test?
MindStudio reported 13.9 tokens per second on GPT-OSS-120B on native Linux on the MSI Prestige RTX Spark laptop. Its Windows result was 65.4 tokens per second.
Did WSL perform better than native Linux?
Yes. The report measured 65.1 tokens per second in WSL, close to its Windows result of 65.4. WSL runs a Linux environment inside Windows and uses the Windows driver stack.
What does the report say causes the slowdown?
MindStudio attributes it to a Windows-only Nvidia power-management component and firmware that Linux cannot read, leaving the GPU near 780 MHz while plugged in in its test. Nvidia’s confirmation of that explanation is not included in the supplied material.
Can Linux run on the RTX Spark laptop?
According to MindStudio, Linux did boot and provide a desktop, CUDA acceleration and other working hardware after configuration. The report describes installation and memory setup obstacles, as well as much lower plugged-in inference performance in its benchmark.
Has Nvidia said when it will fix the issue?
No Nvidia statement or fix timeline appears in the supplied report. MindStudio says a native-Linux improvement would depend on Nvidia adding the relevant power-management logic to its Linux driver.
Source: mindstudio.ai
