#cuda on Hacker News

33 stories tagged cuda, newest first.

Screenshot of infoq.com: TornadoVM 7 Enables GPU-Native Workloads from Java Applications

TornadoVM 7 Enables GPU-Native Workloads from Java Applications (infoq.com)

85 points by pjmlp 4 days ago | 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.

tornadovm java gpu cuda

Screenshot of github.com: Show HN: Rgpu – a PyTorch device whose tensors live on a remote GPU

Show HN: Rgpu – a PyTorch device whose tensors live on a remote GPU (github.com)

46 points by boxstream 19 days ago | 2 comments

GitHub repository for rgpu, a PyTorch device that runs tensor operations on a remote NVIDIA GPU while Python code stays on the local machine, including Macs.

pytorch remote-gpu cuda gpu-computing

Screenshot of dwarfstar.sh: From the creator of Redis; run LLM locally with ds4

From the creator of Redis; run LLM locally with ds4 (dwarfstar.sh)

332 points by fibo 23 days ago | 95 comments

Landing page for DwarfStar 4 (ds4), an MIT-licensed C inference engine for running open-weight LLMs like DeepSeek V4, GLM 5.x and Qwen locally on Mac, CUDA and ROCm.

local-llm-inference ds4 deepseek c

Screenshot of github.com: Show HN: TurboGPT: train 22KiB transformer in 13s

Show HN: TurboGPT: train 22KiB transformer in 13s (github.com)

41 points by lostmsu 26 days ago | 5 comments

GitHub repository for TurboGPT, a CUDA-only project for training a tiny GPT transformer in under a minute, inspired by minGPT.

transformer gpt cuda pytorch

Screenshot of github.com: Show HN: Agentic CUDA Kernel Optimizer

Show HN: Agentic CUDA Kernel Optimizer (github.com)

29 points by bertaye 1 month ago | 6 comments

GitHub repository for a LangGraph-based agentic workflow that uses a C++ CUDA test harness to benchmark, profile, and optimize CUDA kernels.

cuda langgraph ai-agents gpu-optimization

Screenshot of github.com: Show HN: InstinctFlash – Run 5B world-action models in real time on Jetson Thor

Show HN: InstinctFlash – Run 5B world-action models in real time on Jetson Thor (github.com)

17 points by guanming0717 1 month ago | 2 comments

GitHub repository for InstinctFlash, an AGPL-licensed high-performance serving runtime for robotics VLA and world-action models on RTX GPUs and Jetson Thor.

robotics model-inference jetson-thor vision-language-action

Screenshot of bend-lang.com: Bend – A language that blocks AI mistakes via proof, on CPU and GPU

Bend – A language that blocks AI mistakes via proof, on CPU and GPU (bend-lang.com)

564 points by nicolas-siplis 1 month ago | 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.

bend programming-languages formal-verification gpu

Screenshot of developer.nvidia.com: Nvidia announces native GPU programming in Rust

Nvidia announces native GPU programming in Rust (developer.nvidia.com)

898 points by nonmaskable 1 month ago | 354 comments

NVIDIA Technical Blog post introducing CUDA Rust, describing two approaches for writing GPU kernels natively in Rust.

rust cuda nvidia gpu-programming

Screenshot of yassa9.github.io: Geolocating a random island using geometry and CUDA programming

Geolocating a random island using geometry and CUDA programming (yassa9.github.io)

507 points by yassa9 2 months ago | 82 comments

A writeup for the gralhix004 OSINT challenge describing how the author geolocated an islet photo using geometric analysis and CUDA GPU search.

osint geolocation cuda gpu-programming

Screenshot of github.com: Use your Nvidia GPU's VRAM as swap space on Linux

Use your Nvidia GPU's VRAM as swap space on Linux (github.com)

441 points by tanelpoder 4 months ago | 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.

nvidia vram linux swap

Screenshot of nvlabs.github.io: CUDA-oxide: Nvidia's official Rust to CUDA compiler

CUDA-oxide: Nvidia's official Rust to CUDA compiler (nvlabs.github.io)

390 points by adamnemecek 5 months ago | 110 comments

Documentation book for cuda-oxide, an experimental NVIDIA Labs Rust-to-CUDA compiler for writing SIMT GPU kernels in Rust.

rust cuda nvidia gpu-programming

Screenshot of gitlab.com: Nvidia greenboost: transparently extend GPU VRAM using system RAM/NVMe

Nvidia greenboost: transparently extend GPU VRAM using system RAM/NVMe (gitlab.com)

449 points by mmastrac 7 months ago | 126 comments

GitLab repository for nvidia_greenboost, a project to transparently extend Nvidia GPU VRAM using system RAM and NVMe storage.

nvidia gpu vram cuda

Screenshot of github.com: BarraCUDA Open-source CUDA compiler targeting AMD GPUs

BarraCUDA Open-source CUDA compiler targeting AMD GPUs (github.com)

428 points by rurban 8 months ago | 177 comments

GitHub repository for BarraCUDA, an open-source CUDA compiler written in C99 that compiles .cu files to AMD GFX11 GPU machine code.

cuda amd gpu compiler

Screenshot of github.com: Apple's MLX adding CUDA support

Apple's MLX adding CUDA support (github.com)

522 points by nsagent 1 year ago | 181 comments

A GitHub pull request (closed, WIP) in Apple's ml-explore/mlx repository adding an experimental CUDA backend to the MLX framework.

mlx cuda apple nvidia

Screenshot of thenewstack.io: Nvidia adds native Python support to CUDA

Nvidia adds native Python support to CUDA (thenewstack.io)

396 points by apples2apples 1 year ago | 158 comments

The New Stack article reporting that NVIDIA has added native Python support to its CUDA GPU software toolkit.

cuda python nvidia gpu-programming

Screenshot of github.com: DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling (github.com)

381 points by mfiguiere 1 year ago | 65 comments

GitHub repository for DeepSeek's DeepGEMM, a library of clean and efficient FP8 GEMM kernels with fine-grained scaling, released under the MIT license.

deepseek fp8 gemm cuda

Screenshot of github.com: DeepSeek open source DeepEP – library for MoE training and Inference

DeepSeek open source DeepEP – library for MoE training and Inference (github.com)

473 points by helloericsf 1 year ago | 66 comments

GitHub repository for DeepEP, DeepSeek's open-source expert-parallel communication library for Mixture-of-Experts model training and inference.

deepseek mixture-of-experts expert-parallelism gpu-communication

Screenshot of github.com: DeepSeek Open Source FlashMLA – MLA Decoding Kernel for Hopper GPUs

DeepSeek Open Source FlashMLA – MLA Decoding Kernel for Hopper GPUs (github.com)

403 points by helloericsf 1 year ago | 105 comments

GitHub repository for DeepSeek's FlashMLA, an open-source MLA (multi-head latent attention) decoding kernel optimized for NVIDIA Hopper GPUs.

deepseek flashmla multi-head-latent-attention cuda

Screenshot of pyspur.dev: Introduction to CUDA programming for Python developers

Introduction to CUDA programming for Python developers (pyspur.dev)

324 points by t55 1 year ago | 87 comments

A PySpur blog post introducing CUDA GPU programming concepts for Python developers, showing how ML frameworks build on CUDA and its libraries.

cuda gpu-programming python parallel-computing

Screenshot of github.com: CuPy: NumPy and SciPy for GPU

CuPy: NumPy and SciPy for GPU (github.com)

347 points by tanelpoder 2 years ago | 97 comments

GitHub repository for CuPy, a NumPy- and SciPy-compatible array library for GPU-accelerated computing in Python with CUDA and ROCm support.

cupy python gpu numpy

GPU Puzzles (github.com)

345 points by cgadski 2 years ago | 35 comments

GitHub repository by Sasha Rush offering interactive puzzles in a notebook for learning CUDA GPU programming.

cuda gpu-programming puzzles numba

Screenshot of github.com: LibreCUDA – Launch CUDA code on Nvidia GPUs without the proprietary runtime

LibreCUDA – Launch CUDA code on Nvidia GPUs without the proprietary runtime (github.com)

609 points by rrampage 2 years ago | 134 comments

GitHub repository for LibreCUDA, an MIT-licensed project to launch CUDA kernels on Nvidia GPUs without using the proprietary CUDA runtime.

cuda nvidia gpu reverse-engineering

Screenshot of docs.scale-lang.com: Run CUDA, unmodified, on AMD GPUs

Run CUDA, unmodified, on AMD GPUs (docs.scale-lang.com)

1251 points by Straw 2 years ago | 382 comments

Documentation for SCALE by Spectral Compute, a GPGPU toolkit that compiles unmodified CUDA applications natively for AMD GPUs.

cuda amd gpgpu scale

Screenshot of github.com: Nvidia Warp: A Python framework for high performance GPU simulation and graphics

Nvidia Warp: A Python framework for high performance GPU simulation and graphics (github.com)

453 points by jarmitage 2 years ago | 125 comments

GitHub repository for NVIDIA Warp, a Python framework for writing high-performance GPU simulation and graphics code.

nvidia-warp python gpu cuda

Screenshot of github.com: Reproducing GPT-2 in llm.c

Reproducing GPT-2 in llm.c (github.com)

578 points by tosh 2 years ago | 109 comments

A GitHub discussion in Andrej Karpathy's llm.c repository about reproducing GPT-2 training from scratch in C/CUDA.

llm-c gpt-2 karpathy llm-training

Screenshot of hazyresearch.stanford.edu: GPUs Go Brrr

GPUs Go Brrr (hazyresearch.stanford.edu)

1056 points by nmstoker 2 years ago | 243 comments

Stanford Hazy Research blog post on what GPU hardware wants for efficient AI kernels, introducing the ThunderKittens CUDA kernel library.

gpu-kernels thunderkittens cuda flashattention

Screenshot of github.com: Llm.c – LLM training in simple, pure C/CUDA

Llm.c – LLM training in simple, pure C/CUDA (github.com)

810 points by tosh 2 years ago | 134 comments

GitHub repository for llm.c, Andrej Karpathy's project for training LLMs like GPT-2 in simple, raw C/CUDA.

llm gpt-2 c cuda

Screenshot of cgchannel.com: Open-source project ZLUDA lets CUDA apps run on AMD GPUs

Open-source project ZLUDA lets CUDA apps run on AMD GPUs (cgchannel.com)

357 points by drakerossman 2 years ago | 133 comments

CG Channel news article about ZLUDA, an open-source project that lets unmodified CUDA applications run on AMD GPUs.

zluda cuda amd gpu

Screenshot of phoronix.com: AMD funded a drop-in CUDA implementation built on ROCm: It's now open-source

AMD funded a drop-in CUDA implementation built on ROCm: It's now open-source (phoronix.com)

945 points by mfiguiere 2 years ago | 360 comments

Phoronix article reporting that AMD quietly funded ZLUDA, a drop-in CUDA implementation running on ROCm, which has now been released as open source.

zluda cuda rocm amd

Screenshot of codeconfessions.substack.com: What every developer should know about GPU computing

What every developer should know about GPU computing (codeconfessions.substack.com)

422 points by Anon84 3 years ago | 149 comments

A Substack primer by Abhinav Upadhyay explaining GPU architecture and computing concepts for software developers familiar mainly with CPUs.

gpu gpu-architecture parallel-computing cuda

Screenshot of hpcwire.com: AMD may get across the CUDA moat

AMD may get across the CUDA moat (hpcwire.com)

498 points by danzheng 3 years ago | 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.

amd cuda nvidia gpu

Screenshot of github.com: Llama.cpp: Full CUDA GPU Acceleration

Llama.cpp: Full CUDA GPU Acceleration (github.com)

640 points by gzer0 3 years ago | 265 comments

A draft GitHub pull request to llama.cpp adding full CUDA GPU acceleration for all ggml tensors, including storing the KV cache in VRAM.

llama-cpp cuda gpu-acceleration llm-inference

Screenshot of gist.github.com: Run Llama 13B with a 6GB graphics card

Run Llama 13B with a 6GB graphics card (gist.github.com)

564 points by rain1 3 years ago | 233 comments

A GitHub Gist with instructions for running Meta's Llama 13B model on a 6GB GPU using llama.cpp's CUDA/cuBLAS layer offloading.

llama llama-cpp cuda cublas