- ASUS Servers Dominate AI: Advanced Designs Enable Breakthrough Performance Results, Reaching 26 Records Across Six AI Inference Tasks
- ESC8000A-E11 hits 12 records: 4U dual-socket server with simplified thermal design with independent GPU and CPU airflow tunnels
- ESC4000A-E11 achieves 14 records: 2U single-socket server offering a wide range of graphics accelerators, as well as NVIDIA NVLink support
TAIPEI, Taiwan, May 5, 2022 — ASUS, a leading IT company in server systems, server motherboards and workstations, today released its results for the first time since joining MLCommons last December – instantly setting new performance records in dozens of benchmark tasks.
Specifically, in the latest round of MLPerf Inference 2.0, ASUS servers set 26 records in the closed data center division on six AI benchmark tasks, outperforming all other servers with the same GPU configurations. Achievements consist of 12 records achieved with an ASUS ESC8000A-E11 server configured with eight 80GB NVIDIA A100 Tensor Core GPUs; and 14 recordings with an ASUS ESC4000A-E11 server with four 24GB NVIDIA A30 Tensor Core GPUs.
These groundbreaking results clearly demonstrate the dominance of ASUS server performance in the AI realm, bringing significant value to organizations looking to deploy AI and ensuring peak performance in data centers.
ASUS has set 26 AI inference records and tops the scoreboards on six tasks
The MLPerf Inference 2.0 benchmark covers six common AI inference workloads, including image classification (ResNet50), object detection (SSD-ResNet34), medical image segmentation (3D-Unet ), speech recognition (RNN-T), natural language processing (BERT) and recommendation (DLRM).
ESC8000A-E11 has achieved several leading performance positions, including:
- Processing 298,105 frame classifications per second in ResNet50
- Completion of 7462.06 fps object recognition in SSD-ResNet34
- Processing of 24.3 medical images per second in 3D-UNet
- Completed 26,005.7 questions and answers per second in BERT
- Achieved 2,363,760 click-per-second predictions in DLRM
Table of results ESC8000A-E11:
ESC4000A-E11 has achieved several leading performance positions, including:
- Processing 73,814.5 frame classifications per second in ResNet50
- Completion of 1957.18 fps object recognition in SSD-ResNet34
- Processing of 6.83 medical images per second in 3D-UNet
- 27,299.2 speech recognition conversions performed per second in RNNT
- Completed 6,896.01 questions and answers per second in BERT
- 574,371 click predictions made per second in DLRM
Table of results ESC4000A-E11:
Continuous AI performance improvement with optimized server design
The dozen MLPerf Inference 2.0 12 records set by the NVIDIA-certified 4U ESC8000A-E11—configured with eight 80GB NVIDIA A100 PCIe Tensor Core GPUs and two AMD EPYC 7763 processors—demonstrates its supreme scalability for AI and machine learning . Its simplified thermal design, with independent CPU and GPU airflow tunnels, brings a high-efficiency cooling solution to air-cooled data centers.
The NVIDIA-certified ESC4000A-E11, housed in the most compact 2U footprint on the market – and configured with four 24GB NVIDIA A30 PCIe Tensor Core GPUs and two AMD EPYC 7763 processors – established a total of 14 MPLERF Inference 2.0 records . It offers a wide range of graphics accelerators, as well as support for high-speed NVIDIA NVLink GPU interconnect, to unleash maximum AI performance.
Availability and prices
ASUS servers are available worldwide. Please contact your local ASUS representative for more information.
ASUS is a global technology leader that provides the world’s most innovative and intuitive devices, components and solutions to deliver amazing experiences that improve the lives of people everywhere. With its team of 5,000 in-house R&D experts, ASUS is globally known for continuously reinventing today’s technologies for tomorrow, wins more than 11 awards every day for quality, innovation and design, and is ranked among the world’s most admired companies according to Fortune.