68 stories tagged gpu, newest first.
319 points by fragebogen | 74 comments
GitHub repo for a Linux kernel module that unlocks full FP16/BF16 tensor instruction issue rates on Nvidia RTX 3000-series GA10x cards before the Nvidia driver loads.
85 points by pjmlp | 17 comments
InfoQ news article on TornadoVM 7.3, an open-source framework for running Java code on GPUs, covering its FFM backend, CUDA library integration, multi-GPU support and jitLLM.
58 points by ibobev | 2 comments
Catlike Coding tutorial, ninth in a Godot procedural GPU patterns series, showing how to make value noise tile seamlessly in a shader and wrap it around a torus.
41 points by gmays | 5 comments
Blog post by Joe Barrow explaining that speculative decoding speeds up LLM inference by changing the kind of GPU work rather than reducing FLOPs.
308 points by albinstman | 85 comments
Llmspeed is a web tool that simulates the token generation speed of local LLMs on chosen hardware, model, and quantization settings before buying hardware.
60 points by ibobev | 39 comments
A Silicon Underground blog post recounting Nvidia's October 1999 release of the GeForce 256, which it marketed as the first GPU.
7 points by Ir0n | 0 comments
A page from meshapplied.com about running a Qwen LLM at 255 tokens per second on a single Nvidia RTX 5090 GPU; no page content was captured.
239 points by acossta | 96 comments
Apple Machine Learning Research paper page for Minions, a distributed virtual GPU cloud that uses idle Apple Silicon devices for ML inference.
293 points by hurricanepootis | 168 comments
Intel product page for its data center GPU D-series AI inference accelerators, including the D270 with up to 480 GB LPDDR5X memory.
60 points by elffjs | 29 comments
Homepage of Vx, a systems programming language for heterogeneous computing that encodes CPU, GPU, NPU and accelerator memory in its type system.
67 points by luu | 10 comments
GitHub repository for gpuvis, an MIT-licensed GPU trace visualizer tool for viewing GPU and CPU event timelines.
28 points by luu | 3 comments
Blog post by Aras Pranckevičius describing Smol Gaussian, a fast blur technique supporting animated radius, motivated by Blender compositor's slow Fast Blur mode.
92 points by Maverick617 | 17 comments
GitHub repository for Janus, an MIT-licensed AI model API router written in Go with a Vulkan-based runner for GGUF models on AMD, Intel, and Nvidia GPUs.
564 points by nicolas-siplis | 287 comments
Landing page for Bend, a fast programming language with Python-like syntax, CUDA parallelism, and Lean-style proofs to verify AI-written code.
545 points by tolugenius | 385 comments
An Economist briefing arguing that Nvidia acts like a central bank for the AI industry by financing its customers, and asking whether those loans will prove sound.
371 points by simonpure | 244 comments
GPU World is a contest site asking "What if the world had one GPU per person?", with pages for prizes, judges, timeline, guidelines and submissions.
366 points by zhoutong | 88 comments
GitHub repository with patches, tuning files, and config for running DeepSeek V4 Flash on a single AMD MI300X GPU.
441 points by tanelpoder | 111 comments
GitHub repository for nbd-vram, a tool that uses an NVIDIA GPU's VRAM as swap space on Linux via NBD and CUDA, aimed at laptops with soldered RAM.
354 points by sbt567 | 221 comments
Homepage of the Mojo programming language, a Python-like language for fast CPU and GPU code, announcing the 1.0.0b1 stable beta release.
517 points by AbuAssar | 107 comments
Homepage for Lemonade, AMD's open source local LLM server that runs models on GPUs and NPUs on Windows, Linux, and macOS.
562 points by albelfio | 325 comments
Product section of the tinygrad website describing the tinybox, a computer built for deep learning workloads; the page rendered blank when captured.
714 points by mwkaufma | 69 comments
Eric Lengyel's retrospective blog post on ten years of the Slug Algorithm for rendering fonts directly from Bézier curves on the GPU and its licensing history.
449 points by mmastrac | 126 comments
GitLab repository for nvidia_greenboost, a project to transparently extend Nvidia GPU VRAM using system RAM and NVMe storage.
428 points by rurban | 177 comments
GitHub repository for BarraCUDA, an open-source CUDA compiler written in C99 that compiles .cu files to AMD GFX11 GPU machine code.
790 points by ryandrake | 151 comments
Blog post by graphics engineer Sebastian Aaltonen arguing for a simpler, lower-level approach than current graphics APIs, discussing GPU hardware details.
370 points by Garbage | 79 comments
A chapter from Making Software explaining how shaders draw high-fidelity graphics using only x and y pixel coordinates; the capture shows a Vercel security checkpoint.
1204 points by mihau | 1307 comments
Apple press release announcing the M5 chip, with a next-generation GPU featuring per-core Neural Accelerators, a faster CPU and Neural Engine, and higher memory bandwidth.
401 points by Bogdanp | 120 comments
Blog post by Maister describing the design of PyroWave, a custom ultra-low-latency video codec for game streaming.
436 points by birdculture | 252 comments
Granola engineering blog post explaining how a small CSS height animation caused high CPU/GPU usage and how to find and make animations cheaper.
522 points by nsagent | 181 comments
A GitHub pull request (closed, WIP) in Apple's ml-explore/mlx repository adding an experimental CUDA backend to the MLX framework.
392 points by transpute | 80 comments
A personal blog post giving a technical deep dive into Google's TPU architecture, its hardware/software co-design, and how it differs from GPUs.
390 points by ibobev | 121 comments
A blog post by a graphics programmer explaining techniques for rendering crisp, high-quality text on the GPU.
387 points by mfiguiere | 182 comments
Stanford CRFM blog post by Anne Ouyang, Azalia Mirhoseini and Percy Liang describing surprisingly fast AI-generated GPU kernels they discovered.
448 points by rossant | 169 comments
Medium article introducing fastplotlib, a GPU-accelerated interactive plotting library for scientific data visualization.
368 points by ambigious7777 | 403 comments
The Verge news article reporting that power connectors on Nvidia RTX 5090 graphics cards are melting, renewing 12VHPWR cable concerns.
339 points by marban | 203 comments
A personal blog post describing how the author built a budget computer for running AI models locally and privately.
395 points by romes | 192 comments
Inigo Quilez article correcting the misconception that GPU conditionals should be replaced with mix() and step() in shader code.
596 points by eigenvalue | 419 comments
Jeffrey Emanuel's long blog essay arguing the short case for Nvidia stock, citing competition from custom chips and efficiency gains like DeepSeek.
549 points by magicalhippo | 426 comments
TechCrunch article on Nvidia's Project Digits, a compact desktop 'personal AI supercomputer' announced at CES 2025.
399 points by DvdGiessen | 408 comments
Blog post by Matthew Garrett arguing that GPUs, not TPMs, are the real root of hardware DRM; the capture shows a Dreamwidth captcha page instead.
440 points by rbanffy | 632 comments
Phoronix article on Intel's announcement of the Arc B-Series "Battlemage" discrete graphics cards, including the B580, and their Linux support.
560 points by bglazer | 179 comments
Blog post describing a 192-channel open-source phased array microphone with FPGA data acquisition and GPU beamforming and visualization.
347 points by tanelpoder | 97 comments
GitHub repository for CuPy, a NumPy- and SciPy-compatible array library for GPU-accelerated computing in Python with CUDA and ROCm support.
609 points by rrampage | 134 comments
GitHub repository for LibreCUDA, an MIT-licensed project to launch CUDA kernels on Nvidia GPUs without using the proprietary CUDA runtime.
425 points by billconan | 67 comments
WebGPU Unleashed, a free online e-book by Shi Yan teaching graphics programming with WebGPU, with an interactive code playground.
1251 points by Straw | 382 comments
Documentation for SCALE by Spectral Compute, a GPGPU toolkit that compiles unmodified CUDA applications natively for AMD GPUs.
453 points by jarmitage | 125 comments
GitHub repository for NVIDIA Warp, a Python framework for writing high-performance GPU simulation and graphics code.
948 points by LightMachine | 179 comments
GitHub repository for Bend, a massively parallel, high-level programming language from HigherOrderCO that runs on GPUs via the HVM2 runtime.
535 points by seanisom | 151 comments
A former Adobe engineer introduces Renderlet, a WebAssembly-based cross-platform graphics framework, and demos running the Rive 2D renderer on its Wander GPU runtime.
393 points by argulane | 114 comments
Homepage of FuryGpu, a hobbyist project building a fully custom FPGA-based GPU on a PCIe card for modern computers.
385 points by rrampage | 110 comments
Rive blog post announcing that the Rive Renderer, a real-time vector graphics renderer, is now open source and available on all platforms.
339 points by tiahura | 267 comments
CNBC article on Nvidia CEO Jensen Huang unveiling the Blackwell generation of AI GPUs and software, expected to ship later in 2024.
578 points by tosh | 194 comments
Ollama blog post from March 14, 2024 announcing preview support for running models on AMD graphics cards.
821 points by jph00 | 197 comments
Answer.AI blog post announcing an open source system combining FSDP and QLoRA to train a 70B language model on two 24GB consumer GPUs.
357 points by drakerossman | 133 comments
CG Channel news article about ZLUDA, an open-source project that lets unmodified CUDA applications run on AMD GPUs.
727 points by ibobev | 156 comments
Andrew Chan's blog post with notes on the math, algorithms, and methods for simulating fluids like fire and smoke in real time, with source code on GitHub.
336 points by danielhanchen | 107 comments
GitHub repository for Unsloth, an open-source library for faster, lower-memory QLoRA/LoRA finetuning of Llama models using Triton kernels.
997 points by ehPReth | 1053 comments
Apple press release announcing the M3, M3 Pro, and M3 Max chips, its first 3-nanometer chips for personal computers, with a new GPU architecture and faster CPU and Neural Engine.
465 points by superMayo | 65 comments
An interactive blog post, Part I of 'A Journey Into Shaders,' introducing shader programming by building a blob with editable code and sliders.
422 points by Anon84 | 149 comments
A Substack primer by Abhinav Upadhyay explaining GPU architecture and computing concepts for software developers familiar mainly with CPUs.
498 points by danzheng | 266 comments
HPCwire article by Doug Eadline discussing how AMD's Instinct GPUs and software stack might overcome Nvidia's CUDA software lock-in for GenAI workloads.
648 points by birriel | 210 comments
Blog post by Finbarr Timbers explaining the math of why llama.cpp can run LLaMA models locally on consumer hardware, focusing on memory bandwidth and compute limits.
324 points by logicchains | 256 comments
NVIDIA product page for the Grace Hopper Superchip, an accelerated CPU-GPU module for giant-scale AI and HPC applications.
492 points by MacsHeadroom | 328 comments
NVIDIA Technical Blog post announcing the DGX GH200, a Grace Hopper-based system offering 100 terabytes of shared GPU memory for large AI workloads.
365 points by richardliaw | 86 comments
GitHub repository from the Ray project with a cheatsheet of key numbers (costs, token ratios, GPU memory, latency) every LLM developer should know.
608 points by pjmlp | 221 comments
A cohost blog post by mcc (andi) discussing WebGPU, its history, and why it is a compelling graphics API for developers.
377 points by itsuka | 169 comments
WebGPU Fundamentals is a site of tutorial articles teaching WebGPU, covering basics like uniforms, storage buffers, vertex buffers, and textures.
877 points by itsuka | 376 comments
Chrome Developers blog post announcing that WebGPU is shipping in Chrome, enabling modern GPU graphics and compute on the web.