Skip to content

Every morning

GPU news, cut down to what changes your day.

Subscribe for the latest in GPUs, inference engines and accelerator hardware, gathered each morning from the people who build it.

The latest edition

Issue 1, Tue 25 Aug 20265 stories

Hot Chips 2026 Day 2: AMD MI400 and NVIDIA Vera Rubin Take the Stage as llama.cpp and MoE Research Push Inference Forward

Hot Chips 2026's second day delivered architectural deep-dives on AMD's MI400 GPU for Helios racks and NVIDIA's Vera Rubin NVL72, plus Nvidia's unexpected CUDA-on-RISC-V initiative. Meanwhile, llama.cpp shipped per-device Metal flash-attention tuning for Apple Silicon, and a new arXiv framework promises up to 3.1x faster MoE inference on memory-constrained GPUs.

Read the edition

Where it comes from

Read from the people who build it

GPUlse follows the vendors, maintainers and researchers the industry already trusts, and links straight to what they published. No aggregators, and no rewrites of someone else’s summary.

33 sources, read every day

Repositories11
NVIDIA-ISAAC-ROS/isaac_ros_commonNVIDIA/TensorRT-LLMROCm/ROCmflashinfer-ai/flashinferggml-org/llama.cppgoogle-deepmind/mujocohuggingface/lerobotmlcommons/inferencesgl-project/sglangtriton-lang/tritonvllm-project/vllm
Feeds and sites21
arXiv cs.ARarXiv cs.DCarXiv cs.PFchipsandcheese.comdeveloper.nvidia.comhpcwire.commlcommons.orgnewsroom.amd.comnewsroom.arm.comnewsroom.intel.comnextplatform.comphoronix.compr.tsmc.comr/LocalLLaMAr/hardwarerocm.blogs.amd.comsemianalysis.comservethehome.comspectrum.ieee.orgtherobotreport.comtldr.tech
Newsletters by email1
TLDR Hardware

Also from us

GPU CLI, free to use

We also build GPU CLI, a command line tool for running your code on remote GPUs. Put gpu run in front of a command and it executes on rented hardware, with the results synced back to you. Free, and no account needed to start.

gpu-cli.sh