Nvidia says its Blackwell chips lead benchmarks in training AI LLMs
1 min read
Summary
Nvidia unveiled the results of the latest MLPerf Training benchmarking, which the company said demonstrated the success of its next-generation Blackwell chips, which delivered the highest performance of any submission for large language model (LLM) training.
The only submission for each of the MLPerf Training 5.
0 benchmarks, using two AI supercomputers powered by the Nvidia Blackwell platform, underscored the architecture’s versatility and exceptional performance, according to the company.
The chips are being rolled out to AI factories and data centres around the world as part of the company’s aim to speed up the training and deployment of next-gen AI applications.
NVIDIA’s data centre platform includes GPUs, CPUs, fabrics, networking and software, helping businesses to train and deploy models speedily, thus accelerating time to value, the company said.