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 reports that native Linux inference on an MSI Prestige RTX Spark laptop reached 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 a GPU power-management component and encrypted firmware data that Linux cannot currently access; Nvidia has not provided a response in the supplied source material.
A MindStudio report says native Linux inference on an MSI Prestige RTX Spark laptop reached 13.9 tokens per second on GPT-OSS-120B, compared with 65.4 on Windows and 65.1 in Windows Subsystem for Linux (WSL). The report links the performance gap to Linux lacking access to the laptop’s power-management controls, which it says leaves the GPU running at about 780 MHz rather than roughly 2.2 GHz under Windows.
The reported benchmark measured output from GPT-OSS-120B. MindStudio recorded 65.4 tokens per second on Windows, 65.1 in WSL, and 61 on Nvidia’s DGX Spark desktop. Native Linux on the same laptop reached 13.9 tokens per second. Those are results from the report’s test, not a general benchmark covering every RTX Spark model, Linux distribution or workload.
MindStudio says the laptop’s GPU ran near 2.2 GHz on Windows but stayed around 790 MHz during a longer native Linux run. Windows completed that run in about eight minutes, while Linux took 37 minutes. The report says temperatures remained between 40 and 47°C, and attributes the slowdown to power limits rather than heat. It describes the power-table data as encrypted and says the Windows-only Nvidia Platform Controllers and Framework manages CPU and GPU power delivery without a Linux equivalent.
The report also describes setup hurdles separate from inference speed. Linux initially saw only 28GB of the laptop’s 128GB unified memory because firmware reserved most of it for GPU use. After changing the allocation in Windows, Linux could see 120GB, according to MindStudio. With additional configuration and suitable drivers, the laptop reportedly ran a desktop at 4K and 120Hz, with audio, keyboard, touchpad and CUDA acceleration working.
Linux’s GPU Power Limit
The result matters to developers considering RTX Spark laptops for local AI workloads. Many AI tools, including vLLM and SGLang, are commonly used in Linux environments, while these laptops ship with Windows on ARM. The reported measurements suggest that installing Linux directly may sharply reduce inference throughput on the tested machine, even when GPU acceleration is available.
WSL offers a reported workaround: it runs a Linux environment on top of Windows’ driver stack, and its GPT-OSS-120B result was close to native Windows in MindStudio’s test. In a separate vLLM test with eight concurrent clients, the report measured 127 tokens per second in WSL and 39.5 on native Linux. These figures are specific to the reported setup; they do not establish performance across other applications or systems.
The difference could affect whether users choose native Linux, WSL or another machine for development. The report’s findings point to driver and firmware support as the main issue to resolve, rather than a lack of Linux desktop or CUDA functionality alone.
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
Bose QuietComfort Ultra Earbuds (2nd Gen), Wireless Noise-Cancelling Earbuds, Black
As an affiliate, we earn on qualifying purchases.
A Windows-First RTX Platform
The RTX Spark laptop platform is used in systems including models from MSI, Asus and Dell, according to the source report. These machines combine an Nvidia GPU design associated with the DGX Spark desktop with Windows on ARM and unified memory configurations of up to 128GB. The DGX Spark desktop, by contrast, is described as shipping with Nvidia’s own Linux build.
MindStudio’s account concerns an attempted Linux installation on the MSI Prestige N16 Flip AI Plus. It says an initial Omarchy attempt produced an unstable display, while Debian required hardware workarounds. A later Omarchy setup used a kernel with Nvidia’s open driver support, allowing GPU acceleration after the memory allocation was changed in Windows. The report’s broader point is that a working Linux desktop does not by itself guarantee full GPU performance.
The source identifies WSL as the practical option for users who want Linux-oriented AI software while retaining the Windows driver stack. It says WSL maintained GPU boost on both wall power and battery in its testing. The supplied source text cuts off while describing native Linux performance on battery, so it does not establish how the unplugged result compares.
““Windows produced 65.4 tokens per second. WSL … hit 65.1 tokens per second … Native Linux on the laptop … dropped to 13.9.””
— MindStudio report
What the Test Cannot Establish
The supplied material gives no publication date, detailed benchmark methodology, software versions, or independent replication. The reported figures should be read as one test configuration, not a universal measure of RTX Spark performance. Nvidia has not commented in the material provided, and it is unclear whether the company plans Linux support for the platform controller or encrypted power tables.
The source also does not fully explain the security or engineering design behind the encrypted firmware data, nor whether a future driver or firmware update could change the Linux clock behavior. Its account of battery performance is incomplete, ending before giving the result. It is also not clear whether other RTX Spark laptop models behave identically to the MSI system tested.
Nvidia Linux Support Is Key
For now, MindStudio presents WSL as the workaround for users who need Linux-based AI tools while preserving the Windows GPU driver’s reported performance. Users choosing native Linux may need to configure memory allocation in Windows first and should not assume the benchmark results will match their own workload.
The main development to watch is whether Nvidia adds Linux support for the platform’s power-management functions or changes the relevant firmware and driver behavior. The supplied report does not give a company timetable or announce a fix. Further testing across other RTX Spark laptops, workloads and power conditions would help determine how broadly the measured slowdown applies.
Key Questions
How fast was native Linux in the reported test?
MindStudio measured 13.9 tokens per second on GPT-OSS-120B using native Linux on the tested MSI Prestige RTX Spark laptop. The report recorded 65.4 on Windows and 65.1 in WSL.
What does the report say causes the slowdown?
It attributes the gap to Linux lacking access to a Windows-only Nvidia power-management component and to encrypted power-table data. The report says the GPU ran at about 780–790 MHz on Linux, versus roughly 2.2 GHz on Windows.
Does Linux work on the RTX Spark laptop?
According to the report, Linux could boot and run a desktop with GPU acceleration after driver and memory-allocation adjustments. That successful setup did not prevent the reported inference slowdown.
Is WSL faster than native Linux in the report?
Yes. WSL reached 65.1 tokens per second on GPT-OSS-120B, close to Windows at 65.4 and well above native Linux at 13.9 in the reported test.
Has Nvidia announced a fix?
The supplied source material does not report an Nvidia response, fix or release timeline. It says addressing the issue would require Linux support for the platform’s power-management logic.
Source: mindstudio.ai
