Blog
The Linux GPU blog
Practical guides to run your GPU workloads on an integrated GPU — without renting a €300-500/month dedicated GPU.
- GPU VPS comparison
GPU VPS compared: do you really need a datacenter GPU?
Hetzner, OVHcloud and Vultr rent powerful datacenter GPUs, from $0.60/h to over $1,200/month. For transcoding, light AI or headless browsers that is often overkill. Monthly costs compared.
Read the article → - GPU VPS
A cheap GPU VPS for everyday GPU workloads
A flat-price Linux GPU VPS: Intel UHD 770 GPU (Xe-LP, 32 EUs) inside the VM, VAAPI/QuickSync, OpenCL 3.0, Vulkan, OpenVINO. Root SSH, dedicated IPv6, k3s. €14.99/month, no hourly billing.
Read the article → -
Intel Quick Sync (VAAPI) on a Linux VPS: ffmpeg commands that work
Copy-paste ffmpeg VAAPI commands for an Intel UHD 770 on Linux: vainfo, H.264/HEVC encode (classic and low_power), 10-bit HEVC, GPU transcode with scale_vaapi, AV1 decode, GPU monitoring.
Read the article → -
Shared Intel UHD 770 under full load: a 20-minute GPU stress test
Several servers loading the same shared Intel UHD 770 at once for 20 minutes: H.264, 4K HEVC and AV1 transcodes plus Vulkan/OpenCL. 0 failures, 0 GPU hangs, measured throughput per server.
Read the article → -
GPU Cloud: VM or Container, Dedicated or Shared GPU? What Price Comparators Don't Show
GPU price comparators list model, VRAM and price per hour, not whether you get a VM or a container, a whole GPU or a share. Live marketplace data, commands to check from inside, buyer checklist.
Read the article → -
AI Agent on the Web: a Real GPU-Backed Chrome for Claude
Run a Claude agent (Playwright MCP, computer use) in a GPU-backed Chrome on a server: Xvfb vs Xorg dummy, Selenium, Cloudflare bot detection, benchmarks, setup.
Read the article → -
Headless Chrome with GPU on Linux: Playwright, Puppeteer
The right flags to make headless Chrome/Chromium actually use the GPU: ANGLE Vulkan or EGL, verification, Playwright, Puppeteer, Selenium, Docker. 20× faster.
Read the article → -
Jellyfin & Plex Hardware Transcoding: QuickSync on a Rented Server
Enable GPU transcoding (QuickSync / VAAPI) in Jellyfin and Plex on a Linux server with an Intel GPU: Docker, settings, codecs and 4K HDR benchmarks, 5× faster.
Read the article → -
Linux 7.3 and Intel GPUs: What Really Changes (Measured)
Linux 7.3 on an Intel UHD 770 GPU shared between VMs: graphics changes, 7.2 vs 7.3 benchmarks (VAAPI, WebGL, OpenCL), stability with several VMs. Linux GPU servers ship 7.3.
Read the article → -
UHD 770 vs HD Graphics 530: Linux 3D and transcoding benchmarks
Measured Intel UHD 770 and HD Graphics 530 results on Linux: glmark2, H.264 encoding, and two simultaneous 4K transcodes, including test limits.
Read the article → -
Run Hugging Face Models on an Integrated GPU (iGPU) — Ollama & llama.cpp
Pull open models from Hugging Face and run them on an integrated GPU (iGPU) via Ollama and llama.cpp (Vulkan): small LLMs like Gemma, Phi, Qwen. Private, cheap inference from €14.99/month.
Read the article → -
GPU-Accelerated Selenium & Playwright: No €500/Month GPU Required
Run your Selenium and Playwright tests with GPU acceleration (WebGL, Canvas, video) on a server with an integrated iGPU/APU. Fast, true-to-life browser-test CI from €14.99/month, k3s/k8s.
Read the article → -
Affordable AI Inference: OpenVINO on an Integrated GPU, No €500/Month Card
Serve your AI models (vision, OCR, lightweight LLMs) with accelerated inference on an Intel iGPU via OpenVINO / ONNX Runtime. Low latency, low cost, from €14.99/month, k3s/k8s.
Read the article → -
Headless 3D Rendering (Blender EEVEE, OpenGL/Vulkan) Without a €500/Month GPU
Render your scenes and images server-side on an integrated GPU: Blender EEVEE, OpenGL/Vulkan engines, image generation. From €14.99/month, k3s/k8s.
Read the article → -
Xvfb: Practical Guide and How to Get Real GPU Rendering
One-line xvfb-run for Playwright, Selenium and Puppeteer. Why Xvfb renders on the CPU (llvmpipe) and how to make Chrome use the GPU, with measurements.
Read the article → -
Julia Accelerated with OpenCL on an Integrated GPU, No €500/Month Card
Run your Julia jobs and notebooks on an integrated-GPU server (Intel UHD 770) via OpenCL: linear algebra, simulation, data. From €14.99/month, root SSH, k3s/k8s.
Read the article → -
k3s / Kubernetes Clusters with Integrated GPUs, No €500/Month Nodes
Deploy your containerized GPU workloads on a k3s / Kubernetes cluster where every node carries an integrated GPU (iGPU/APU), exposed via a device plugin. From €14.99/month, k3s option +5 €/node.
Read the article → -
Hardware-Accelerated Video Transcoding: You Don't Need a €500/Month GPU
Encode and decode video with hardware acceleration (VAAPI, Intel QuickSync) on an integrated-GPU server from €14.99/month. FFmpeg, H.264/H.265, streaming, k3s/k8s.
Read the article →