Insights on AI inference benchmarking, GPU performance, and ML infrastructure.
New to the terminology? Browse the AI inference glossary.
Rubin LUT Based Tensor Core, Feynman, Rack Scale, Perf Per MegaWatt, Perf Per Dollar, Software Improvements, Public Rubin Software, PyTorch, vLLM, OpenAI Triton
Day 0 Inference Performance, InferenceX, 100x performance improvement in 26 Days, Cost per Million Tokens, Huawei 950DT Inference Trace Analysis
DSv4-Pro FP4 8K/1K, Dynamo+vLLM, disaggregated on both racks. GB300's 50% extra HBM (288 vs 192 GB/GPU) unlocks a wider prefill+decode recipe GB200 can't fit — lifting middle-of-curve perf/$ by 2.31x despite a 20% per-GPU TCO premium.
The amd/deepseek_v4 side branch shipped TileLang attention indexer, Triton sparse MLA, fused RoPE/Hadamard, FlyDSL MoE, and FP4 weights across 31 performance optimizations PRs — lifting first-light 20 tok/s/GPU at 2.4 tok/s/user into 2,256 tok/s/GPU at 9.4 tok/s/user on 8K/1K, with both throughput and interactivity climbing together
DeepSeek R1 FP4 1k/1k. NVL72's 72-GPU NVLink scale-up fabric lets decode run wide EP up to EP=32, where B200's 8-GPU NVLink island caps out at EP=8 over RoCEv2
Piecewise CUDA graphs for DeepSeek V3, a unified event loop, and JIT kernels push 8k/1k throughput from 508 to 907 tok/s/GPU on the same 16 GPU B200 pool
