NVIDIA Breaks AI Inference Records with New A30 and A10 Server GPUs
NVIDIA achieves top performance across all MLPerf benchmarks while introducing the A30 and A10 GPUs for mainstream enterprise servers.
NVIDIA sets new AI inference records across every category in the latest MLPerf benchmarks. The company expands its hardware lineup with the introduction of the NVIDIA A30 and A10 GPUs, which are designed specifically for mainstream enterprise servers. These new chips combine high performance with low power consumption to handle a wide variety of AI workloads.
The A30 and A10 GPUs target enterprise applications that require efficient AI inference, training, and graphics capabilities. Major server manufacturers including Cisco, Dell Technologies, Hewlett Packard Enterprise, Inspur, and Lenovo plan to integrate these new GPUs into their highest-volume server models starting this summer. This broad adoption brings powerful AI computing to standard data center infrastructure.
NVIDIA stands as the only company to submit results for every test in both the data center and edge categories of the MLPerf benchmark. The company leverages its complete AI software stack, including TensorRT and Triton Inference Server, alongside the NVIDIA Ampere architecture's Multi-Instance GPU capabilities to achieve these industry-leading results.