InferenceX
Named by 14 kept items in this workspace.
13 items name it under frameworks and tools and 1 under classification.key entities.
Search every workspace for InferenceX. It matches the name as written, so it can find other things that share it.
The items that name it 14
Independent inference benchmark from SemiAnalysis; DeepInfer is absent from it.
Inference benchmarking suite on which DeepInfer cannot be compared because of its offline, single-turn limit.
Public benchmark results and recipes cited by vLLM and SemiAnalysis.
Accelerator benchmark used for TPUv7, GB300 and Vera Rubin submissions.
Accelerator benchmark used for TPUv7 vs GB300 submissions.
SemiAnalysis benchmark; added B200 configurations.
SemiAnalysis benchmark; B200 configurations added
Inference benchmark with preview submission cited by SemiAnalysis
Realtime benchmark board used to compare modeled serving cost across GPUs, with an ownership cost model at hyperscaler scale.
Open-source inference benchmark from SemiAnalysis, used here for agentic inference serving results on Engram models across NVIDIA GPUs and MI355X.
Open-source inference benchmark and data covering TPUv7, NVIDIA and AMD, with SambaNova and Trainium to follow. PowerX integration is planned for finer throughput per gigawatt measurements.
SemiAnalysis inference benchmark, used to run Kimi K3 on 16 GB300 GPUs (8 prefill, 8 decode) with HBM utilization cut from 92% to 85%.
SemiAnalysis benchmark used here for the first third-party TPUv7 Ironwood inference results against Nvidia B200/B300.
All technologies in this workspace.