
The latest MLPerf Inference 4.1 benchmark results have been released, showcasing significant advancements in AI hardware and software. Nvidia's new Blackwell GPU made its debut, demonstrating up to four times the performance of its predecessor, the H100, in certain tests, including Llama 2 70B using FP4. This marks a notable achievement for Nvidia, as the Blackwell series sets a new standard for generative AI systems. The benchmarks highlight the rapid pace of innovation in AI technologies, with contributions from other major players including AMD, Google, and Untether AI. The results indicate that AI inference capabilities continue to improve, driven by both hardware advancements and software optimizations.



Learn more about our #MLPerf #Inference v4.1 results in this terrific writeup by @TechJournalist and @VentureBeat 🙏 https://t.co/93DY0D9YfT
Thank you to @muradhem and @the_logic for the write-up on our recent #MLPerf #Inference v4.1 results. https://t.co/VAES5OArxU
NVIDIA Blackwell Joins MLPerf Inference with Upgraded Performance Metrics https://t.co/KEQb8AWWld @nvidia #datanami #TCIwire