Microsoft's Azure AI Supercomputer delivers record MLPerf benchmarks

Recently, Microsoft Azure joined the Top 10 club of the the TOP500 super computer rankings by delivering 30.05 Petaflops. It was based on Microsoft’s recently announced Azure NDm A100 80GB v4, available on demand. These Azure NDm A100 v4 instances are powered by NVIDIA GPU acceleration and NVIDIA InfiniBand networking.

Microsoft today highlighted the latest (December 2021) MLPerf 1.1 results in which Azure delivered the #2 performance overall and the #1 performance by a cloud provider.

1. Approximately 25 seconds to train BERT Large natural language processing model on 2,048 GPUs
2. Processed up to 3.8M images per sec using Resnet50 1.5 image classification with 2,048 GPUs
3. Completed the Minigo (reinforcement learning) benchmark in under 17.5 minutes with 1,792 GPUs.

The above results showcase Azure’s commitment to scale and performance for AI training using cloud computing.

The latest MLPerf Training v1.1 benchmark round received submissions from 14 organizations and released over 185 peer-reviewed results for machine learning systems spanning from edge devices to data center servers. Submissions this round included software and hardware innovations from Azure, Baidu, Dell, Fujitsu, GIGABYTE, Google, Graphcore, HabanaLabs, HPE, Inspur, Lenovo, NVIDIA, Samsung, and Supermicro. You can find the full results here.

Source: Microsoft