Headless DGX Spark: Freeing Up 2 GB Display Reserved Memory in UEFI

X postIvan FioravantiTutorial

Ivan Fioravanti explains how to reclaim system memory on a headless DGX Spark. By default, the system reserves 2 GB of RAM for the display, a value that can now be reduced in recent BIOS versions.

Key points

  1. DGX Spark reserves 2 GB of system RAM for display output by default.
  2. On headless deployments, this memory can be freed via the UEFI setup menu under Advanced settings.
  3. The setting requires a modern BIOS version, and the author recommends running a system update beforehand.
  4. The workaround originates from NVIDIA documentation, whereas community members note that some third-party variants (such as ASUS or MSI) still lack lower memory allocations.

Why it matters

For memory-intensive AI workloads, every gigabyte of system RAM counts. On headless nodes where no physical display is attached, reserving memory for video output is an unnecessary overhead.

Analysis

This memory allocation setting was unavailable in earlier BIOS revisions of the DGX Spark. Fioravanti notes that applying the latest system updates is essential before accessing the UEFI menu. The thread also sparked discussion regarding remote configuration via SSH and differing BIOS options across OEM platforms.

Original post

Ivan Fioravanti

@ivanfioravanti · 2 November 2026

On a new DGX Spark that you use headless 👇 Change Display Reserved Memory DGX Spark reserves 2 GB of system memory for the display by default. Free them! 1. Restart DGX Spark. 2. Immediately press Esc or Del and hold it until the UEFI setup menu appears. 2. Go to Advanced ->

213 likes28 replies272 bookmarks13.3k views

View on X
More posts in the thread (9)
  1. @15ky3 Nuuuu, I bet they'll add it. NVIDIA needs a BIOS upgrade to get this choice.
  2. @neojohnson113 Nuuuu, I bet they'll add it. NVIDIA needs a BIOS upgrade to get this choice.
  3. Before doing this, be sure to do a system update.
  4. @ajshpprd 8 is a lot! I discovered this recently. It was not available in earlier versions of the BIOS.
  5. I should use AI to review my posts. I manually edited this from NVIDIA manual, removed 2 steps and made a mess with numbering 😂
  6. @robotnurseio Yes, it should stick.
  7. Does anyone know if this is feasible on MSI?
  8. @michaeljonathan You are right, I tried to slim down everything possible, but not easy there.
  9. @tomiw83 I bet they'll add it! /cc @rldfst90
Selected replies (5)
  • @BTCXoomer @ivanfioravanti Do it from the command line, no keyboard, no monitor needed. Give this to your agent: # Free ~1.5 GB of RAM on a headless DGX Spark — no monitor or BIOS visit needed Every DGX Spark's UEFI carves **2 GB of system RAM** out for the display before Linux ever boots. That memory is
  • @15ky3 @ivanfioravanti Not possible on Asus gx10 😥 There is only the Option for 2 and 4 GB
  • @neojohnson113 @ivanfioravanti Tried this on my Asus gx10 with latest update, but no 0.5GB yet in BIOS
  • @ajshpprd @ivanfioravanti Times 8 that's 12 GB I didn't know I was missing 👀
  • @XadenRyan @ivanfioravanti @TechMDAI With the new firmware you can do this over ssh now

Summary generated by AI (Gemini 3.8 Flash, 4 November 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • X post:Mia

    Running DeepSeek v4 Flash on a Single NVIDIA DGX Spark

    MiaAI-Lab published an open-source launcher allowing DeepSeek V4 Flash 0731 (EXL3) to run on a single NVIDIA DGX Spark node instead of requiring a multi-node setup.

    71bookmarks9.8kviews

    AI & LLMs· Tool

  • Repository:MiaAI-Lab/DeepSeek-v4-Flash-One-DGX-Spark

    Running DeepSeek-v4-Flash on a Single DGX Spark

    An open-source launcher and Docker recipe for running DeepSeek V4 Flash 0731 (EXL3) on a single NVIDIA DGX Spark with 128 GiB unified memory.

    328starsPython

    AI & LLMs· Tool

  • Article:Elite Test Engineering

    Serving DeepSeek V4 Flash on Dual DGX Spark Nodes

    A detailed technical setup report documents serving DeepSeek-V4-Flash (284B MoE) across two DGX Spark nodes powered by Grace Blackwell chips using Ray and vLLM. It highlights workarounds for unified memory monitoring, NCCL RDMA deadlocks, and CUDA graph issues, as well as a 45% speedup using native speculative decoding.

    AI & LLMs· Tutorial

  • X post:Mia

    NVIDIA Sync: Local Network Access and Setup for DGX Spark

    An NVIDIA guide demonstrates how the NVIDIA Sync desktop application streamlines connecting laptops to DGX Spark devices over a local network, replacing manual terminal commands.

    160bookmarks9kviews

    AI & LLMs· Tutorial

  • Link:Donato Capitella

    Strix Halo AI Toolboxes: Containerised Environments and Host Tuning for AMD Ryzen AI MAX+

    Donato Capitella provides Strix Halo AI Toolboxes, a collection of containerised environments and host configuration guides for AMD's Ryzen AI MAX+ platform. Leveraging unified memory, the hardware allows the integrated GPU to address up to 128 GB of RAM for local LLMs, image generation, and model training.

    AI & LLMs· Resource

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.