gpu
84 stories and discussions about gpu, aggregated from every source we track.
The real difference between texture-atlas, SDF, MSDF, and Slug glyph rendering on the GPU: how each works, where each breaks, and when to use which.
[Experimental] A virtio device for near-native NVIDIA GPU access in KVM virtual machines. - nestrilabs/virtio-nvgpu
The GPU in your computer is about 10 to 100 times more powerful than the CPU, depending on workload. For real-time graphics rendering and machine learning, you are enjoying that power, and doing those workloads on a…
A high-performance and high-level purely functional data-parallel array programming language that can execute on the GPU and CPU.
There's a new kind of technical debt, and it doesn't come from cutting corners. It comes from...
Serving Gemma 4 E2B q4_0 through llama.cpp on one laptop, twice: CPU-only and on a 2021-era 4 GB GTX 1650 Ti. Same GGUF, same binary, same prompts, one flag apart. The card takes decode by 4.3x, and needs only 1598 MiB to do it.
Single-source pure-Rust heterogeneous GPU compute platform - enkiruntime/enki
Gemma 4 E2B q4_0 served by llama.cpp on one laptop, CPU-only and on a GTX 1650 Ti, rebuilt on CUDA 13.4 and re-measured in CPU, GPU, GPU, CPU order with a temperature gate. The card takes decode by 4.14x, and run order moves the answer by about 2%.
Step-by-step: running Google's quantization-aware-trained Gemma 4 E2B on a 10th-gen Core i7 laptop with a 4 GB GTX 1650 Ti — why bf16 and int8 cannot fit, why the QAT GGUF does with room to spare, and managing it with an MCP server.
The live demo, step by step: Gemma 4 E2B answering at 76 tok/s from a GTX 1650 Ti with 4 GB of memory, and the exact re-pack of Google's QAT weights on Hugging Face that makes it fit, run faster and stay close to bf16.
Apple's container CLI 1.4.1 cannot boot the official debian:13 image as a machine, because it has no /sbin/init. A small Dockerfile fixes that and gives you a persistent Debian 13 VM with systemd, your Mac user and your home folder. The VM has no GPU, so Ollama runs on macOS and the VM calls it at 192.168.64.1:8000. From inside Debian, gemma4:e2b answered at 45.2 tok/s, loaded 100% on the M3 GPU.
Pretrained text classifiers you can run and retrain on CPU - nicobrenner/jeffy
How our Kimi K3 megakernel on TPU v7 reaches over 700 tokens/s with speculative decoding and nearly 2× GB200's batch-one decode throughput.
Maximizing perf on AI-SQL queries with the KV-optimal left-deep join
Nobody ever thinks of RISC-V when they think of Nvidia, but it's actually incredibly important to Nvidia's GPUs.
Bulk Synchronous Parallelism (BSP) in distributed deep learning clusters induces severe rate-of-change current transients (dI/dt) across data center power delivery networks. When thousands of accelerators synchronously…
A curated resource list for learning AI performance engineering, from GPU fundamentals to production inference. - wafer-ai/gpu-perf-engineering-resources
[Experimental] A virtio device for near-native NVIDIA GPU access in KVM virtual machines. - nestrilabs/virtio-nvgpu
A GPU-accelerated Pedersen vector commitment engine for Ethereum Verkle trees (EIP-6800) - Dyslex7c/cuda-verkle
A homelab census of 41 Docker and 9 LXC containers, and how one 6 GB GPU decides which AI agents run on local Ollama, OpenRouter or a subscription.
We introduce PTXBench, a benchmark for evaluating and adapting large language models (LLMs) to use architecture-specific PTX for GPU kernel optimization. PTXBench measures functional correctness, whether selected…
GPU Trace Visualizer. Contribute to mikesart/gpuvis development by creating an account on GitHub.
KernelBench: Can LLMs Write GPU Kernels? - Benchmark + Toolkit with Torch -> CUDA (+ more DSLs) - ScalingIntelligence/KernelBench
GPUs are expensive, yet inference-serving GPU clusters remain heavily underutilized. To improve utilization, state-of-the-art systems adopt GPU multiplexing. However, optimizing solely for utilization can…
Crop a line of text and get the closest fonts from Google Fonts, DaFont, Adobe Fonts, GitHub font repositories, Debian and more, with their weight and italic.
Glyd stores open models' bf16 weights in about 11 bits instead of 16 and decodes them inside the GPU's matrix multiply: bit for bit, often faster.
VRAM-aware single-node GPU job scheduler. Contribute to ceruleane/gpusched development by creating an account on GitHub.
NEO Emacs (WIP): GPU powered Emacs written in Rust with a modern display engine. Aiming for modern design & multi-threaded Elisp, 10x performance, zero-pause concurrent GC and 100% Emacs compat...
A high-performance, GGUF-native Rust & CUDA inference engine optimized for cold-start latency and real-time 'System 1' agent decision loops. - lateos-ai/reflex
We won 2nd overall at HackMIT for building a GPU in 24 hours.
A modern LLM can spend most of its time doing something that looks almost embarrassingly simple: C...
A high-performance and high-level purely functional data-parallel array programming language that can execute on the GPU and CPU.
You need to first ensure that all the necessary drivers (CUDA most important) for the Nvidia GPU are installed. This post will NOT cover how to do...
DevOps & Cloud SI Assistant that runs locally with Ollama for complete privacy. - tara-vision/taracode
GPU-Rendered UI library built from scratch in Rust [WIP] - artemtsitronov/glacex
Five years of GPU infrastructure at Beam — from ECS and Knative cold starts to a custom container runtime, FUSE lazy-loading, and a trustless binary.
Stop wasting hours waiting for video to be built from the web page in a browser. Vibe code your video on GPU using Rust & SVG framework.
Institutional-grade real-time AI cloud GPU spot rates, cluster arbitrage, and inventory intelligence across 31 global cloud providers.