» Tag
cuda
14 postsNew Verifier Exposes Hidden Bugs in LLM-Generated GPU Kernels
A 12-gate contract verifier finds that 39.5% of LLM-generated GPU kernels are broken despite passing standard correctness benchmarks.
CUDA 13.3 Brings Hardware Carryless Multiplication to GPUs
CUDA 13.3's new clmad PTX instruction accelerates carryless multiplication on GPUs, delivering up to 18.8x faster GHASH and 4-13x faster sum-check.
Tesla P100's silent FP16 precision bug in llama.cpp fixed
A 3-line CUDA patch in llama.cpp fixes a years-old silent FP16 precision bug on Tesla P100 GPUs, with zero performance cost.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comPXQ format revives Pascal/Volta GPUs: +88% prefill on a $150 Tesla P100
PXQ format and custom CUDA kernels in ik_llama.cpp fork pxq_llama deliver major prefill/decode speedups on Tesla P100, V100 and GTX 1080 Ti GPUs.
Flash-MSA: Open-Source Sparse Attention Kernels for Fast Training
Flash-MSA brings open-source CuTeDSL training kernels for MiniMax Sparse Attention on Hopper/Blackwell GPUs, with linear-time backward passes.
Nunchaku Lite Brings 4-bit Diffusion Inference Natively to Diffusers
Nunchaku's 4-bit SVDQuant inference is now natively loadable in Diffusers via from_pretrained, no separate engine or local CUDA compile needed.
CAKE: Compiler-Agent Co-Design Lets AI Write Faster GPU Kernels
CAKE pairs AI coding agents with a hardware-explicit compiler IR, beating tuned GPU kernel baselines on B200 with verifiable, evolving toolchains.
Bw24: From-Scratch Rust+CUDA LLM Engine Tuned for RTX 50 Blackwell
Bw24 is a from-scratch Rust+CUDA LLM inference engine tuned for RTX 50-series Blackwell GPUs, hitting up to 2.3x speculative decoding gains over llama.cpp.
The Cost of Irregularity: CUDA C++, Rust, and Triton
A comparison of CUDA C++, Rust, and Triton in GPU programming. Performance differences in irregular workloads are analyzed.
From CUDA to MLX: K-Search Optimizes for Apple Silicon
K-Search translates CUDA optimizations into MLX for Apple Silicon, marking a key advancement for GPU kernels in AI.