Can you run Linux on the Nvidia RTX Spark laptop?
Yes, but it comes alongside a grave catch. Linux installs and runs on RTX Spark laptops akin the MSI Prestige N16 Flip AI Plus, including GPU acceleration through Nvidia’s open kernel driver. The issue is clock speed: formerly the laptop is plugged into partition power, the GPU gets locked to approximately 780 MHz on Linux alternatively of the 2 GHz-plus it hits on Windows. That turns a part capable of about 65 tokens per second on a 120 milliard indicator example into one struggling to interrupt 14 tokens per second, for reasons tied to power administration firmware that Linux currently can’t read.
TL;DR
- Dual-booting Linux works on the MSI Prestige RTX Spark laptop, alongside a complete desktop, GPU acceleration via Nvidia’s open driver, 4K output at 120Hz, and operating audio, keyboard, and touchpad formerly configured correctly.
- A hidden Windows-only component called the Nvidia Platform Controllers and Framework manages power shipment between CPU and GPU, and it has no Linux equivalent, which is the base logic of the slowdown.
- GPU clock speed collapses to concerning 780 MHz on Linux whenever plugged in, compared to approximately 2.2 GHz on Windows, since the firmware’s power array is encrypted and lone the Windows controller can decrypt it.
- Benchmarks on GPT-OSS-120B display the gap clearly: Windows hit 65.4 tokens per second, WSL hit 65.1, the Nvidia DGX Spark hit 61, and native Linux on the identical laptop managed lone 13.9.
- Memory allocation is a distinct headache, since the laptop reserves 96GB of unified recollection for the GPU by default and lone Windows exposes a environment to alter that split, which is required before Linux can see additional than a sliver of RAM.
- WSL (Windows Subsystem for Linux) is the applicable center dirt today, since it rides on the Windows controller stack and keeps complete GPU boost whether the laptop is plugged in or on battery, during motionless operating Linux-first AI tools akin vLLM.
- A genuine fix depends on Nvidia, specifically on porting the phase controller’s power administration logic into its Linux driver, which would efficiently rotate the laptop into a portable type of the DGX Spark.
Self-host
Remy.
On-prem or in your personal cloud. Remy and all app you build remain inner your walls.
What makes the RTX Spark laptop distinct from a normal Windows PC?
The RTX Spark platform, which ships in machines akin the MSI Prestige, the Asus ProArt, and the Dell XPS, uses the identical GPU scheme as Nvidia’s DGX Spark desktop unit. But anywhere the DGX Spark ships alongside Nvidia’s own Linux build out of the box, all one of these laptop variants ships alongside Windows on ARM.
That difference matters since Windows on ARM alongside CUDA assistance is new. ARM-based Windows machines have existed for a few years, but pairing that architecture alongside Nvidia’s CUDA stack lone became imaginable alongside this chip. The laptops additionally assistance unified recollection configurations up to 128GB, shared between CPU and GPU, which is different for a laptop and chief to why they’re being marketed at AI developers in the archetypal place.
Because so much of the AI tooling ecosystem, including vLLM and SGLang, is built Linux-first, developers who buy these machines naturally desire to cognize whether they can ditch Windows, or at smallest dual-boot, without losing performance.
How do you really instal Linux on it?
The procedure is not plug-and-play. A archetypal attempt using Omarchy (a Linux allocation alongside Arch roots) unsuccessful because, during it already supported the part on Asus and Dell hardware, the MSI Prestige behaved differently, producing a flickering, unstable display during boot.
Switching to Debian worked additional reliably, but motionless needed workarounds. The installer couldn’t nexus to Wi-Fi, so facility required a USB Ethernet adapter. Out of the box, Debian additionally didn’t acknowledge the keyboard, touchpad, or GPU, since none of the laptop-specific hardware assistance was current in a vanilla kernel.
Going rear to Omarchy solved the controller problem, since its kernel already bundled Nvidia’s open-source driver. At that item Linux booted properly, but could lone see 28GB of the laptop’s 128GB of unified memory. That’s since the laptop’s firmware reserves 96GB for GPU use, and that divided is lone adjustable from Windows. Dropping the Windows-side GPU recollection booking downward to 4GB freed up 120GB for Linux to see formerly booted. After that adjustment, the laptop ran a complete desktop at 4K and 120Hz, alongside operating audio, touchpad, keyboard, and CUDA acceleration.
Getting video output complete HDMI during the instal procedure was its own obstacle. Early in the process, the laptop would lone display video on its own inner screen, not complete HDMI, which required pointing an external camera at the laptop’s display fair to see what was happening during remote-controlled facility steps.
How does Linux achievement difference to Windows and WSL?
This is anywhere the narrative turns. With GPT-OSS-120B operating on the GPU, Windows produced 65.4 tokens per second. WSL, which runs a Linux surroundings inner Windows, hit 65.1 tokens per second, nearly identical. The DGX Spark desktop unit, operating Nvidia’s native Linux build, came in at 61 tokens per second. Native Linux on the laptop, however, dropped to 13.9 tokens per second, approximately five times slower.
Every team builds alongside AI.
Nothing gets lost.
The gap showed up again alongside vLLM operating eight concurrent clients: WSL reached 127 tokens per second during native Linux reached 39.5. A longer benchmark run told the identical narrative in period fairly than throughput. Windows completed in concerning 8 minutes during holding the GPU near 2.2 GHz. Linux took 37 minutes, alongside the GPU parked about 790 MHz the complete time. Temperatures stayed low, between 40 and 47°C, so this isn’t a thermal throttling issue. The GPU merely isn’t being allowed to run at complete speed.
Oddly, the direction of the issue flips depending on power source. Unplugged, Linux hit about 61 to 67 tokens per second on LM Studio, comparable to Windows. Plugged in, that identical test dropped to concerning 15 tokens per second. Clock speed told the identical story: about 1,800 MHz unplugged under load, and downward to approximately 780 MHz whenever connected to partition power.
Why does plugging in the laptop dilatory Linux down?
The logic traces rear to a component apparent in Windows Device Manager called the Nvidia Platform Controllers and Framework. This part of firmware and controller logic manages the shared power prosperity between CPU and GPU, since a laptop (unlike the desktop DGX Spark) has to balance power shipment between the two chips fairly than feeding the GPU on its own line.
Disabling that controller in Windows reproduced the slowdown directly: tokens per second dropped from about 83 to concerning 31, and clock speed cut from approximately 2 GHz downward to the identical 786 MHz seen on Linux whenever plugged in. Re-enabling the controller restored normal speed immediately. That confirms the controller, not the kernel, the distribution, or the Nvidia GPU controller itself, is what’s holding Linux back.
The deeper matter is that the firmware hands this controller a array of power settings that is encrypted. Windows has the key to decrypt it. Linux currently does not, and no application fix attempted during evaluation could allow it, including one attempt that triggered a kernel panic. Extracting the decryption key from the Windows controller was floated as a theoretical workaround, but that would amount to a fragile, single-machine hack fairly than item that plant broadly. A genuine fix needs to arrive from Nvidia shipping equal power administration logic in its Linux driver.
Is WSL a fine substitute for native Linux correct now?
For most developers, yes, at smallest for now. WSL runs on top of the Windows controller stack, so it inherits Windows’ complete GPU boost behavior whether the laptop is plugged in or operating on battery. It additionally runs Linux-first AI tools akin vLLM and SGLang without the clock throttling issue that affects native Linux installs.
The one adaptation needed is memory. WSL reserves lone 14GB by default, which isn’t adequate for ample example workloads on a 128GB unified recollection machine, so that allocation needs to be risen manually. Once that’s done, WSL benchmarks district near to native Windows performance, making it the additional applicable choice until Nvidia addresses the Linux controller gap directly.
Frequently Asked Questions
Can the Nvidia RTX Spark laptop dual-boot Windows and Linux?
Yes. Dual-booting was demonstrated on the MSI Prestige RTX Spark laptop using Omarchy, a Linux allocation that already included Nvidia’s open kernel controller for this chip, following an first attempt alongside Debian required manual controller and networking workarounds.
Why is Linux so much slower than Windows on this laptop whenever plugged in?
Developers have their tools.Enterprises have Remy.
Developers
> add charge limiting to the API
● Reading src/server.ts
● Editing 3 files
{ bounds }
from 'rate'
- app.use(cors())
✓ Tests passed · prepared for PR
Enterprises
Loan Approvals● Live
SOC 2Role-based accessSelf-hosted
Mortgage · $412,000Approved
Small endeavor · $85,000Approved
Auto · $31,500Approved
A Windows-only firmware component called the Nvidia Platform Controllers and Framework manages power sharing between the CPU and GPU. Without a Linux equivalent, the GPU’s clock speed gets locked about 780 MHz whenever the laptop is connected to power, compared to approximately 2.2 GHz on Windows.
Does unplugging the laptop fix the Linux achievement problem?
It helps significantly. On power division power, Linux GPU clocks and token throughput arrive much nearer to Windows levels. The serious slowdown specifically appears whenever the laptop is plugged into partition power, which is the contrary of what most users would expect.
Is WSL the identical as operating native Linux on the RTX Spark?
Not exactly, but it’s near in practice. WSL runs a Linux surroundings on top of the Windows kernel and controller stack, so it retains Windows’ GPU clock boost behavior and avoids the throttling issue, during motionless supporting Linux-native AI tools akin vLLM.
Will Nvidia fix Linux GPU achievement on RTX Spark laptops?
That depends on Nvidia shipping the phase controller’s power administration logic in its Linux driver. The encrypted firmware power array currently can lone be decrypted by Windows, and no workaround established so far restores complete clock speed on Linux without it.
