Ollama AMD GPU on Windows — Custom Build (680M/780M/890M)

Run Ollama with full AMD GPU acceleration on Windows — even on unsupported integrated GPUs like the 680M, 780M, and 890M. This step-by-step guide shows you how to build a custom Ollama installer with HIP SDK 6.4.2 for complete GPU offloading on AMD Radeon iGPUs. Products in this video (affiliate links support the channel at no extra cost to you): - GMKtec M7 Mini PC (Radeon 680M): https://www.amazon.com/dp/B0DQKRXKS5?tag=hakehardware-20 - GMKtec K12 Mini PC (Radeon 780M): https://www.amazon.com/dp/B0FK2BLV7J?tag=hakehardware-20 - GMKtec EVO-X2 AI Mini PC (128GB RAM): https://www.amazon.com/dp/B0F53MLYQ6?tag=hakehardware-20 - Samsung 990 EVO Plus 4TB SSD: https://www.amazon.com/dp/B0DHLBDSP7?tag=hakehardware-20 0:00 - Intro 0:46 - GPU Support Overview (680M, 780M, 890M) 1:52 - Required Tools Overview 2:20 - Install Go & MinGW 3:49 - Set Environment Variables 4:40 - Install AMD Drivers & Git 6:07 - Install HIP SDK 6.4.2 7:32 - Replace ROCm Library Files 10:42 - Install Build Tools & CMake 13:20 - Clone & Configure the Ollama Repo 16:57 - Run the Build Script 18:01 - Install Your Custom Ollama Build 19:26 - Start Ollama & Verify GPU Detection 20:47 - Pull & Run a Model with Full GPU Offload 27:59 - Wrap Up Related Videos: - ComfyUI on AMD 780M with Zluda: https://youtu.be/LVjOoTJ4NiQ - Local AI with a Mini PC: https://youtu.be/ve4pR7OrEsQ - Strix Halo + ROCm 7.1 + Ubuntu: https://youtu.be/vX7cYahvcPI - ComfyUI + ROCm 7.1 + Strix Halo: https://youtu.be/8a_oANfzQ-8 - The Best Way to Run Local AI (LM Studio): https://youtu.be/0S_uIqdehEo Required Downloads: - Go: https://go.dev/dl/ - MinGW: https://github.com/niXman/mingw-builds-binaries/releases - AMD Adrenaline Drivers: https://www.amd.com/en/support/download/drivers.html - Git: https://git-scm.com/install/windows - HIP SDK 6.4.2: https://www.amd.com/en/developer/resources/rocm-hub/hip-sdk.html - ROCm Libraries (680M/780M): https://github.com/likelovewant/ROCmLibs-for-gfx1103-AMD780M-APU/releases - Build Tools: https://visualstudio.microsoft.com/visual-cpp-build-tools/ - CMake: https://cmake.org/download/ - Inno Setup: https://jrsoftware.org/isdl.php Find Your GPU Code: - AMD GPU Codenames: https://github.com/likelovewant/ollama-for-amd/wiki/AMD-GPU-Codenames - TechPowerUp GPU Database: https://www.techpowerup.com Written Guide: https://localhake.com/content/ollama-for-amd-windows #Ollama #AMD #LocalAI #ROCm #Windows