Artificial intelligence benchmarking organization MLCommons released a new set of tests and results on Wednesday that evaluate how quickly top hardware can run artificial intelligence applications and respond to users.
Two new benchmarks added by MLCommons measure how quickly AI chips and systems can generate responses from powerful AI models filled with data. The results give a rough idea of how quickly AI applications like ChatGPT can respond to user queries.
One of the new benchmarks adds the ability to measure the speed of large language models in question-answering scenarios. It’s called Llama 2, contains 70 billion parameters, and was developed by Meta Platforms.
MLCommons officials have also added a second text-to-image generator to the benchmark tool suite, called MLPerf, based on Stability AI’s Stable Diffusion XL model.
Servers powered by the Nvidia H100 chip, made by companies including Alphabet’s Google, Supermicro and Nvidia, handily won two new raw performance benchmarks. Several server makers have submitted designs based on the company’s less powerful L40S chip.
Server maker Krai submitted a baseline design for image generation using Qualcomm’s AI chip, which consumes far less power than Nvidia’s cutting-edge processors.
Intel also submitted a design based on its Gaudi2 accelerator chip. The company described the results as “robust.”
Raw performance isn’t the only important metric when deploying AI applications. Advanced AI chips consume large amounts of energy, and one of the most significant challenges facing AI companies is deploying chips that provide maximum performance with the least amount of energy.
MLCommons has a separate benchmark class for measuring power consumption.
© Thomson Reuters 2024
(This story has not been edited by NDTV staff and is auto-generated from a syndicated feed.)
Follow us on Google news ,Twitter , and Join Whatsapp Group of thelocalreport.in
