Pure Triton kernels for Qwen3.5-27B inference on NVIDIA B200

A USRP B200 compatible GPSDO board with the u-blox LEA-M8F

Compare H100, H200, B200, MI300X TCO across cloud providers and on-prem — open-source MCP agent demo.

GPU kernel lab for NVIDIA Blackwell (B200) — run CUDA, CuTe, CUTLASS, Triton, CuTe DSL, and Quack kernels from a single CLI using Modal.

Spheron Network is a decentralized GPU and cloud compute marketplace that aggregates enterprise-grade NVIDIA GPU capacity from certified Tier 3/4 data centers worldwide and exposes it through a single on-demand, per-minute billed interface. The marketplace covers H100, H200, B200, B300, A100, GH200, L40S, RTX PRO 6000, RTX 5090, RTX 4090, and…

Code for pre-training a GPT-2 model on (eight) NVIDIA DGX B200 GPUs and short tutorial on the topic. Uses Torch and HF Transformers. It can pre-train GPT-2 Small on 32 GB of data in around 2.5 hours. It handles dataset tokenization too.

Run BEVFormer (ECCV 2022) on PyTorch 2.8 + Blackwell B200; SimpleDDP fix for mmcv 1.x on PyTorch 2.x + Tensor Core analysis