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

Displays drive information in the order they are inserted in the bays of your rack mounted storage

Thermal-aware batch controller for vLLM/TensorRT-LLM. Prevents HBM thermal throttling from killing p99 latency on H100/H200. Monitors nvidia-smi, auto-cuts batch size at 85°C, migrates cold KV to DRAM. Prometheus + Grafana included. 4.2s -> 2.1s p99 at 128K context.

Monitor low-utilization time, idle-state episodes, and workload starvation signals on NVIDIA datacenter GPUs.

Daily-verified cloud GPU rental prices across 18+ providers (H100, H200, B200, A100, RTX 4090...). Open dataset, updated daily, CC BY 4.0. Live at gpurentalprices.com

GLM-5.2 744B at 4-bit on Modal 4x H200 via vLLM, plus a static streaming chat UI.

dd-ready fully packed FreeDOS disk image for crossfalshing SAS2008 like Dell H200 to HBA

AMD MI300X vs NVIDIA H200: a controlled single-GPU benchmark of inference and training. AMD's 1.32x paper FLOPs advantage inverts on achieved efficiency, but its 1.36x memory becomes a 1.84x KV-cache advantage that wins large-model serving outright.

⚡ TIMTEH Model Forge — Uncensored, abliterated & reasoning-distilled GGUFs. Forged on 8×H200 SXM5 | 1.1TB VRAM

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

SGPU — Simple GPU monitor for the SGVR H200 lab (MLXP / NAVER Cloud NKS). Zero-install kubectl dashboards, per-pod GPU process attribution, in-pod TUI, and per-user usage accounting.