Nvidia's new llm benchmark tool fixes the benchmark, not the model. aiperf, out september 18,…
nvidia's new llm benchmark tool fixes the benchmark, not the model. aiperf, out september 18, replaces genai-perf with a multiprocess design so the client stops bottlenecking high-concurrency tests, and it supports 15+ endpoint types.
measuring inference was the problem.
Context
The ai-dynamo/aiperf repository describes AIPerf as a benchmarking tool for generative AI model performance with latency metrics. Quantum Zeitgeist, 19 September 2026, says it is a ground-up rewrite not built on Perf Analyzer, that GenAI-Perf was single-process and constrained by the Python GIL under realistic loads, that AIPerf supports over fifteen endpoint types, GPU telemetry via DCGM or pynvml and TTFT, ITL and throughput with percentiles. A migrations doc snippet calls it a drop-in replacement for GenAI-Perf with a cut-off qualifier.
The multiprocess design and 15 plus endpoint types are secondary only; first-party architecture text was not read. Out September 18 has no first-party release date, and the secondary pieces are dated 19 September. Replaces is not confirmed, since the migration doc's qualifier was cut off and deprecation of GenAI-Perf was not seen. The note's framing that it fixes the benchmark and not the model is the author's.
Watch next
- NVIDIA's announcement and the GenAI-Perf feature comparison page.
Sources
- ai-dynamo/aiperf (GitHub)github.com
- NVIDIA AIPerf to reliably test LLM (Quantum Zeitgeist, 19 Sep 2026)quantumzeitgeist.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 21 September 2026 at 01:05 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →