NVIDIA Outperforms the Latest Artificial Intelligence Benchmarking Tests

In the third round of submission, MLCommons recently released the results for MLPerf Inference. It is a set of standard AI inference benchmarking tests using seven different applications. These seven tests are predominantly based on the range of workloads that include computer vision, medical imaging, recommender systems, and natural language processing. The test mainly measures how fast the neural network can process data from each application.

In the results announced, the NVIDIA A100 GPU was the highest performing accelerator in each application. The company submitted each and every offline and server data scenario. It outperformed the much-impressive Qualcomm AI 100 in every aspect and also by an extensive margin. NVIDIA's edge GPUs - A100 PCIe and A100 SXM-was also introduced. The company submitted Jetson Xavier NX and Jetson AGX Xavier with a higher performance than the NX version. 

NVIDIA's all-new A100 has 80 GB of memory, and it provides over two terabytes per second of memory bandwidth, which is the highest for any AI-based company. With enhanced power, it is known that the applications cannot fully utilize A100 capabilities. It separated the GPU into seven isolated instances to create the biggest MIG. Along with Qualcomm, Centaur also had submissions. Again, NVIDIA dominated the performance landscape in every case.