# \#ai **URL:** https://community.frame.work/tag/ai/205.md [Latest](https://community.frame.work/latest.md) · [Categories](https://community.frame.work/categories.md) · [Tags](https://community.frame.work/tags.md) --- ## [hipEngine: Fast Native Qwen 3.6 Inference for RDNA3 (including Strix Halo)](https://community.frame.work/t/hipengine-fast-native-qwen-3-6-inference-for-rdna3-including-strix-halo/82803)
**Author:** [@lhl](https://community.frame.work/u/lhl)\ **Replies:** 19\ **Last updated:** [October 23, 2026, 5:28pm UTC](https://community.frame.work/t/hipengine-fast-native-qwen-3-6-inference-for-rdna3-including-strix-halo/82803 "2026-10-23T17:28:00Z")
Hey everyone, I wanted to share GitHub - shisa-ai/hipEngine · GitHub - a new open source (from-scratch) inference engine that is a pure AMD ROCm implementation (all hot-path code is HIP/C++, all AMD libraries, PyTorch is… --- ## [Local LLM Ops Dashboard / Proxy](https://community.frame.work/t/local-llm-ops-dashboard-proxy/85569)
**Author:** [@ndom91](https://community.frame.work/u/ndom91)\ **Replies:** 0\ **Last updated:** [October 20, 2026, 1:57pm UTC](https://community.frame.work/t/local-llm-ops-dashboard-proxy/85569 "2026-10-20T13:57:01Z")
Hey folks, I’ve spent the last few months working on this self-hosted ops dashboard for local LLMs. It’s generic enough to be used anywhere, but I’ve gotten a lot out of it on my framework mainboard-based system. https:… --- ## [Gufo: the all-in-one strix halo inference engine](https://community.frame.work/t/gufo-the-all-in-one-strix-halo-inference-engine/85083)
**Author:** [@Federico\_Izzo](https://community.frame.work/u/Federico_Izzo)\ **Replies:** 13\ **Last updated:** [October 14, 2026, 10:24pm UTC](https://community.frame.work/t/gufo-the-all-in-one-strix-halo-inference-engine/85083 "2026-10-14T22:24:15Z")
Premise: cross-post from reddit. Hello, we are happy to share a project that a friend of mine and I have been working on for a while. It’s an extremely optimized inference engine for the Strix Halo platform. We both hav… --- ## [Bluefin and Desktop TPM Conflicts](https://community.frame.work/t/bluefin-and-desktop-tpm-conflicts/85260)
**Author:** [@rlathastak](https://community.frame.work/u/rlathastak)\ **Replies:** 4\ **Last updated:** [October 5, 2026, 2:35am UTC](https://community.frame.work/t/bluefin-and-desktop-tpm-conflicts/85260 "2026-10-05T02:35:59Z")
I am trying to set up a Desktop (395) that I just got. I am struggling to get the Bluefin OS to “play” nice when TPM is turned on. I need an immutable Linux OS with native container/Docker support that can run AI, and B… --- ## [FW16 newer Radeon GPU module needed for actual NPU+GPU hybrid work via Lemonade](https://community.frame.work/t/fw16-newer-radeon-gpu-module-needed-for-actual-npu-gpu-hybrid-work-via-lemonade/83682)
**Author:** [@Jen\_Savage](https://community.frame.work/u/Jen_Savage)\ **Replies:** 12\ **Last updated:** [September 29, 2026, 11:51pm UTC](https://community.frame.work/t/fw16-newer-radeon-gpu-module-needed-for-actual-npu-gpu-hybrid-work-via-lemonade/83682 "2026-09-29T23:51:20Z")
I have been exploring the features of my FW16 (AI 9 HX 370) and it turns out that AMD’s local‑AI software stack (Lemonade) only reaches its full performance when both the CPU/NPU and GPU are AMD. So my fancy shmancy nvid… --- ## [Setup recommendations for AI running](https://community.frame.work/t/setup-recommendations-for-ai-running/84823)
**Author:** [@Windmill\_tilter](https://community.frame.work/u/Windmill_tilter)\ **Replies:** 9\ **Last updated:** [September 23, 2026, 12:03pm UTC](https://community.frame.work/t/setup-recommendations-for-ai-running/84823 "2026-09-23T12:03:01Z")
I am purchasing a Desktop with Max+ 395 - 64GB to run local AI. I will use voice to text (e.g. whisper), documents (RAG) and general AI prompts. I have experience with Ubuntu and Arch (CatchyOS), also Ollama in the cont… --- ## [Framework Camera Tuning](https://community.frame.work/t/framework-camera-tuning/84455)
**Author:** [@Sh3rl0cked](https://community.frame.work/u/Sh3rl0cked)\ **Replies:** 3\ **Last updated:** [August 29, 2026, 1:09pm UTC](https://community.frame.work/t/framework-camera-tuning/84455 "2026-08-29T13:09:43Z")
Hey all! Love my Framework 13 Pro, and I noticed the camera was terrible in video calls. My job and personal life call for a lot of video calls so this wasn’t acceptable to me. I spent some time (and some tokens) to bu… --- ## [Slow LLM Performance on Framework 13 Pro (358H)](https://community.frame.work/t/slow-llm-performance-on-framework-13-pro-358h/84361)
**Author:** [@Mmmmm](https://community.frame.work/u/Mmmmm)\ **Replies:** 5\ **Last updated:** [August 26, 2026, 6:09pm UTC](https://community.frame.work/t/slow-llm-performance-on-framework-13-pro-358h/84361 "2026-08-26T18:09:07Z")
I just received my Framework 13 Pro with the X7 358H with 32gb Crucial LP-CAMM2 7500Mhz in Omarchy 4. I just tried running a simple prompt with a lm-studio and gemma-4-12b-qat (a 7gb model). I was a bit surprised that it… --- ## [Question about Framework Desktop 192 GB as a local AI workstation (Strix Halo + dGPU)](https://community.frame.work/t/question-about-framework-desktop-192-gb-as-a-local-ai-workstation-strix-halo-dgpu/83865)
**Author:** [@tommaso\_di\_leo](https://community.frame.work/u/tommaso_di_leo)\ **Replies:** 3\ **Last updated:** [August 4, 2026, 4:14pm UTC](https://community.frame.work/t/question-about-framework-desktop-192-gb-as-a-local-ai-workstation-strix-halo-dgpu/83865 "2026-08-04T16:14:10Z")
Hi everyone, I’m very interested in the upcoming Framework Desktop 192 GB configuration, especially for running large local language models such as DeepSeek V4 Flash. I’ve been following the recent work around DwarfSta… --- ## [When might we see a beefier Framework laptop?](https://community.frame.work/t/when-might-we-see-a-beefier-framework-laptop/83282)
**Author:** [@moreplz](https://community.frame.work/u/moreplz)\ **Replies:** 10\ **Last updated:** [July 31, 2026, 12:43pm UTC](https://community.frame.work/t/when-might-we-see-a-beefier-framework-laptop/83282 "2026-07-31T12:43:24Z")
The announcement of the FW13 Pro is awesome. I’m really glad to see the company really dialing in a sleek and light option for professionals that’ll give them Frameworks legendary modularity and repairability. But now I… --- ## [PCIe USB4 expansion card for AI desktop](https://community.frame.work/t/pcie-usb4-expansion-card-for-ai-desktop/83327)
**Author:** [@David\_McKinnon](https://community.frame.work/u/David_McKinnon)\ **Replies:** 3\ **Last updated:** [July 2, 2026, 4:37pm UTC](https://community.frame.work/t/pcie-usb4-expansion-card-for-ai-desktop/83327 "2026-07-02T16:37:06Z")
Haven’t found answers to this yet searchign the forums - I’m trying to chain a number of Framework desktops together for distributed AI processing, but this is limited by only having two USB4 slots in the back. Are there… --- ## [Agentic AI box](https://community.frame.work/t/agentic-ai-box/83177)
**Author:** [@John\_Hartemink](https://community.frame.work/u/John_Hartemink)\ **Replies:** 2\ **Last updated:** [June 23, 2026, 11:01pm UTC](https://community.frame.work/t/agentic-ai-box/83177 "2026-06-23T23:01:05Z")
Has anyone had any experience running agentic AI on a dedicated framework mainboard? I have moved mine into the coolermaster case (16Gb DDR5, Intel-ultra-1) and I am looking for some more interesting things than just a m… --- ## [\[Solved\] How do I increase GTT on Arch Linux with Bootctl?](https://community.frame.work/t/solved-how-do-i-increase-gtt-on-arch-linux-with-bootctl/81037)
**Author:** [@LunarLattice](https://community.frame.work/u/LunarLattice)\ **Replies:** 4\ **Last updated:** [June 18, 2026, 7:33pm UTC](https://community.frame.work/t/solved-how-do-i-increase-gtt-on-arch-linux-with-bootctl/81037 "2026-06-18T19:33:50Z")
I feel like I’ve tried almost anything that I’ve managed to google, but nothing seem to work so far. Followed the installation for Arch, installed KDE. amdgpu\_top shows 64042 MiB GTT available. As following the instal… --- ## [Feedback for plan: cluster of 4 framework motherboards with RoCE netorking cards](https://community.frame.work/t/feedback-for-plan-cluster-of-4-framework-motherboards-with-roce-netorking-cards/81475)
**Author:** [@Kristof\_Csillag](https://community.frame.work/u/Kristof_Csillag)\ **Replies:** 4\ **Last updated:** [May 11, 2026, 11:07pm UTC](https://community.frame.work/t/feedback-for-plan-cluster-of-4-framework-motherboards-with-roce-netorking-cards/81475 "2026-05-11T23:07:37Z")
Hi all, I am planning the build an AI cluster from 4 Frameboard motherboards in a DeskPi RackMate T1 case. Similar to what Jeff Geerling has done, with the exception that I want way faster interconnect, so I want to add… --- ## [\[GUIDE\] Use NPU (XDNA2) with Arch Linux and FastFlowLM!](https://community.frame.work/t/guide-use-npu-xdna2-with-arch-linux-and-fastflowlm/80879)
**Author:** [@GreyXor](https://community.frame.work/u/GreyXor)\ **Replies:** 11\ **Last updated:** [April 27, 2026, 12:25pm UTC](https://community.frame.work/t/guide-use-npu-xdna2-with-arch-linux-and-fastflowlm/80879 "2026-04-27T12:25:19Z")
Good News, Everyone! We can now use the NPU for LLM inference with FastFlowLM. Here’s my little guide for you to achieve it To use the NPU on Arch Linux you need some patches that will soon be merged into linux, xrt… --- ## [Optimal LLM Inference Configuration for Ryzen AI MAX+ 395 on Proxmox with GPU Passthrough — Seeking Community Input](https://community.frame.work/t/optimal-llm-inference-configuration-for-ryzen-ai-max-395-on-proxmox-with-gpu-passthrough-seeking-community-input/81467)
**Author:** [@Wolfgang](https://community.frame.work/u/Wolfgang)\ **Replies:** 3\ **Last updated:** [April 26, 2026, 4:50pm UTC](https://community.frame.work/t/optimal-llm-inference-configuration-for-ryzen-ai-max-395-on-proxmox-with-gpu-passthrough-seeking-community-input/81467 "2026-04-26T16:50:40Z")
Proxmox + Ryzen AI MAX+ 395: Optimizing LLM Inference with GPU Passthrough Hi together, We run a 2-node Proxmox cluster with Ryzen AI MAX+ 395 CPUs (124 GB RAM each) for self-hosted LLM inference. GPU passthrough (VFIO)… --- ## [Bigger Keyboard Request](https://community.frame.work/t/bigger-keyboard-request/81791)
**Author:** [@mike\_t](https://community.frame.work/u/mike_t)\ **Replies:** 1\ **Last updated:** [April 13, 2026, 6:20am UTC](https://community.frame.work/t/bigger-keyboard-request/81791 "2026-04-13T06:20:58Z")
Hi There Framework Team, I’m new purchaser, here are my specs System: AMD Ryzen™ AI 300 Series - Ryzen™ AI 9 HX 370 Expansion Bay Module: Graphics Module (AMD Radeon™ RX 7700S) (2nd Gen) Expansion Bay: Expansion Bay … --- ## [Framework Desktop Gen AI Build April 2026 (Ubuntu 24.04.4 + ROCm + llama.cpp)](https://community.frame.work/t/framework-desktop-gen-ai-build-april-2026-ubuntu-24-04-4-rocm-llama-cpp/81759)
**Author:** [@Matthew\_Richardson](https://community.frame.work/u/Matthew_Richardson)\ **Replies:** 1\ **Last updated:** [April 11, 2026, 12:47pm UTC](https://community.frame.work/t/framework-desktop-gen-ai-build-april-2026-ubuntu-24-04-4-rocm-llama-cpp/81759 "2026-04-11T12:47:31Z")
Just to be clear, this is not spam, just the steps I took to get my desktop up and running with the stated stack. After some trial and error, I was able to get Ubuntu 24.04.4 installed with ROCm and llama.cpp running. F… --- ## [Step-by-step guide: Ubuntu 24.04 + ROCm 7.2.1 + llama.cpp on Framework Desktop](https://community.frame.work/t/step-by-step-guide-ubuntu-24-04-rocm-7-2-1-llama-cpp-on-framework-desktop/81721)
**Author:** [@Inndy](https://community.frame.work/u/Inndy)\ **Replies:** 2\ **Last updated:** [April 11, 2026, 8:25pm UTC](https://community.frame.work/t/step-by-step-guide-ubuntu-24-04-rocm-7-2-1-llama-cpp-on-framework-desktop/81721 "2026-04-11T20:25:55Z")
Hello folks, If you’ve been wanting to run LLMs locally on your Framework Desktop but weren’t sure where to start — I put together a walkthrough that covers the full setup from scratch. The guide walks through: UEFI … --- ## [London users/user groups?](https://community.frame.work/t/london-users-user-groups/81751)
**Author:** [@megs-p](https://community.frame.work/u/megs-p)\ **Replies:** 1\ **Last updated:** [April 10, 2026, 4:06pm UTC](https://community.frame.work/t/london-users-user-groups/81751 "2026-04-10T16:06:39Z")
Are there Framework enthusiasts in London? would love to know your experience so far! --- ## [Confused Flux 2 Dev - 130gb Overload](https://community.frame.work/t/confused-flux-2-dev-130gb-overload/81583)
**Author:** [@Dunc\_Ward1](https://community.frame.work/u/Dunc_Ward1)\ **Replies:** 2\ **Last updated:** [April 2, 2026, 7:26am UTC](https://community.frame.work/t/confused-flux-2-dev-130gb-overload/81583 "2026-04-02T07:26:57Z")
’m very new to Framework, and I’m also new to APUs/iGPUs, so I’d appreciate some straightforward advice. I’ve got a Framework Desktop Max+ 395 with 128GB RAM on the way. Specs: 16 cores / 32 threads 3.0GHz base c… --- ## [Performance observation (with fix / workaround): FL13 Intel 11th Gen + AMD AI Pro R9700 + Razer Core X + LM Studio](https://community.frame.work/t/performance-observation-with-fix-workaround-fl13-intel-11th-gen-amd-ai-pro-r9700-razer-core-x-lm-studio/81527)
**Author:** [@Second\_Coming](https://community.frame.work/u/Second_Coming)\ **Replies:** 0\ **Last updated:** [March 25, 2026, 11:52pm UTC](https://community.frame.work/t/performance-observation-with-fix-workaround-fl13-intel-11th-gen-amd-ai-pro-r9700-razer-core-x-lm-studio/81527 "2026-03-25T23:52:31Z")
Background: For whatever reason, the R9700 in the Core X would perform very poorly in LM Studio. Software: Windows 11 25H2 AMD Adrenaline 26.3.1 LM Studio 0.47 + Vulkan runtime 2.8.0 + Qwen 3.5 35B A3B Q4\_K\_M Hardw… --- ## [Anyone using Proxmox VE?](https://community.frame.work/t/anyone-using-proxmox-ve/74863)
**Author:** [@Bradley\_G](https://community.frame.work/u/Bradley_G)\ **Replies:** 35\ **Last updated:** [March 21, 2026, 5:29pm UTC](https://community.frame.work/t/anyone-using-proxmox-ve/74863 "2026-03-21T17:29:45Z")
Hey desktop crew. I wanna use this thing for multiple use-cases simultaneously. Namely, game servers, AI API server (ollama or vllm) and various other web servers maybe in podman containers. So, Proxmox seems ideal. Now … --- ## [\[Projects\] Nyx — opinionated NixOS + Niri setup for Framework laptops (desktop router concept)](https://community.frame.work/t/projects-nyx-opinionated-nixos-niri-setup-for-framework-laptops-desktop-router-concept/81342)
**Author:** [@Ethan\_Kelley](https://community.frame.work/u/Ethan_Kelley)\ **Replies:** 1\ **Last updated:** [March 20, 2026, 6:53am UTC](https://community.frame.work/t/projects-nyx-opinionated-nixos-niri-setup-for-framework-laptops-desktop-router-concept/81342 "2026-03-20T06:53:01Z")
Hi everyone :waving\_hand: I wanted to share a project I’ve been building around running NixOS on the Framework 13 AMD (Strix Point / AI 300) platform. The project is called Nyx, and the goal is not just to maintain dot… --- ## [Updated commands to increase max unified memory usage on Framework Desktop under Fedora 43?](https://community.frame.work/t/updated-commands-to-increase-max-unified-memory-usage-on-framework-desktop-under-fedora-43/78460)
**Author:** [@milkmandan](https://community.frame.work/u/milkmandan)\ **Replies:** 24\ **Last updated:** [March 14, 2026, 11:24pm UTC](https://community.frame.work/t/updated-commands-to-increase-max-unified-memory-usage-on-framework-desktop-under-fedora-43/78460 "2026-03-14T23:24:09Z")
Been looking high and low for the proper commands to run under a fresh Fedora 43 install for Framework Desktop on how to increase the VRAM allocation for running larger LLM models under LM Studio. I have the desktop conf… --- ## [How does LInux use the NPU](https://community.frame.work/t/how-does-linux-use-the-npu/79889)
**Author:** [@DexxFM](https://community.frame.work/u/DexxFM)\ **Replies:** 6\ **Last updated:** [February 27, 2026, 9:00am UTC](https://community.frame.work/t/how-does-linux-use-the-npu/79889 "2026-02-27T09:00:15Z")
I’m a recent (as of like 30 mins ago) purchaser of a Framework 13! I was curious about how the built in NPU works with non-Windows OSes. How does Linux use the NPU in the AMD AI 3k series APU? I know windows has some bu… --- ## [Anyone utilized the NPU?](https://community.frame.work/t/anyone-utilized-the-npu/77846)
**Author:** [@John\_Myers](https://community.frame.work/u/John_Myers)\ **Replies:** 11\ **Last updated:** [February 27, 2026, 8:59am UTC](https://community.frame.work/t/anyone-utilized-the-npu/77846 "2026-02-27T08:59:48Z")
I’m trying to work with the NPU on linux but running ‘xrt-smi examine’ shows zero devices found. Is there a bios setting I’m not seeing that needs to be changed to enable the NPU? I see the NPU in lspci, the firmware i… --- ## [Claw Command @ CSULA | Project Discussion](https://community.frame.work/t/claw-command-csula-project-discussion/80897)
**Author:** [@Claw\_Command\_Cyber\_C](https://community.frame.work/u/Claw_Command_Cyber_C)\ **Replies:** 2\ **Last updated:** [February 26, 2026, 5:56am UTC](https://community.frame.work/t/claw-command-csula-project-discussion/80897 "2026-02-26T05:56:01Z")
New Univ. Club, Fresh Hardware, and a Lot of Curiosity Well met, Framework Community! We’re Claw Command, a brand new cybersecurity club at Cal State LA. Our unofficial motto is \*\*\*"do cool sh\*\*\*\*t" which we think can l… --- ## [Local GitHub Copilot with Lemonade Server on Windows](https://community.frame.work/t/local-github-copilot-with-lemonade-server-on-windows/80851)
**Author:** [@Guest574](https://community.frame.work/u/Guest574)\ **Replies:** 0\ **Last updated:** [February 24, 2026, 12:31pm UTC](https://community.frame.work/t/local-github-copilot-with-lemonade-server-on-windows/80851 "2026-02-24T12:31:55Z")
When I got my Framework desktop one of the things I wanted to try was using it for local firstcoding assistance. It took a while for me to work out how get that up and running so I wrote the guide that I wanted on how to… --- ## [FW13 AMD "Reverse Charging" loop with high-capacity Power Banks](https://community.frame.work/t/fw13-amd-reverse-charging-loop-with-high-capacity-power-banks/80848)
**Author:** [@Imran\_Bin\_ilias](https://community.frame.work/u/Imran_Bin_ilias)\ **Replies:** 0\ **Last updated:** [February 24, 2026, 11:55am UTC](https://community.frame.work/t/fw13-amd-reverse-charging-loop-with-high-capacity-power-banks/80848 "2026-02-24T11:55:21Z")
Hey everyone, I’m seeing a weird behavior with my FW13 AMD (7840U) on BIOS 3.05 when using high-wattage power banks like the Anker 737 or the UGreen Nexode 200W. Instead of the laptop “sinking” power from the bank, it s… [Next page](https://community.frame.work/tag/ai/205.md?match_all_tags=true&page=1&tags%5B%5D=ai)