Bit-exact emulation of NVIDIA tensor-core FP8 GEMMs (torch._scaled_mm, Transformer Engine) as an H100 PCIe or RTX 4060 computes them, on the CPU or another CUDA GPU
Communication-layer expert-parallel MoE dispatch/combine optimization on H100 with PyTorch all-to-all and DeepEP
Host-side scheduler and transfer-path optimization in SGLang on H100 with C++ descriptors, Triton and pinned memory
DeepSeek-V3, R1 671B on 8xH100 Throughput Benchmarks
From-scratch, heavily-annotated CUDA inference runtime for Qwen2.5-Coder-7B on H100 (sm_90). Custom INT4 packer, fused GEMV, paged KV, split-KV attention, CUDA graph decode — every hot path commented for the why. Educational, not a llama.cpp replacement.
Multi-GPU tensor-parallel vLLM on AMD Radeon RX 7900 XT / XTX / GRE (7900XT, 7900XTX, RX7900XT, gfx1100, RDNA3, ROCm): root cause and fix for the RCCL hostcall / PCIe atomics (AtomicOps) crash "NCCL error: unhandled cuda error" / "operation cannot be performed in the present state", Proxmox VFIO passthrough; LLM inference benchmarks on 13 machines
Dashboard for AI Studio, Open Source Continuous Inference | Deepseek-R1, Qwen2.5, Llama3.1 | 4xRTX-5090 inside PRU2500, 2xH100 inside PRU2500, 8xMI210 in SuperMicro
LLM benchmarking, GPU workload orchestration backend server | Deepseek-R1, Qwen2.5, Llama3.1 | 4xRTX-5090 inside PRU2500, 2xH100 inside PRU2500, 8xMI210 in SuperMicro
Sub-real-time MiniMax H3 on Hopper: 13.506 s for a 14.375 s 768p video with audio on 8xH100. Four runtime patches removing 9.72 s of non-model overhead from FastVideo.
A Flexible and High-Performance Inference Serving Engine for Open Diffusion Language Models
Rent ready-to-use cloud GPUs in seconds. Lium CLI makes it easy to launch, manage, and scale GPU compute directly from your terminal. Fast, cost-optimized, and built for AI & ML developers.
Fast linear-attention, SSM and attention kernels (Pallas Mosaic GPU) and a training stack for JAX on NVIDIA Hopper
CLI tool to check Oracle Cloud compute shape availability - find capacity across regions
Faster attention kernels for serving TML's Inkling model on vLLM. 2.7x over the shipping path on H100, and the only implementation that runs on A100.
Bunch of explanations and tutorials around confidential computing
A single executable notebook showing how to use turbopairformer, developed by Lambda, Stevens Institute of Technology, and OpenFold — standalone CUDA kernels for the two most expensive operations in an AlphaFold3-style pair stack.
PyTorch DDP and multi-GPU LLM training — from a minimal distributed example to NanoGPT speedruns, Muon, H100 profiling, and modded-nanogpt. FBA LAB https://bubblnet.com
To associate your repository with the h100 topic, visit your repo's landing page and select "manage topics."