Saturday, August 8, 2026
DarkSubscribe
AI Infrastructure · News & Analysis
HomeChips & HardwareReport
Chips & Hardware · Report

NEUCHIPS' RecAccel N3000 GPU achieved competitive inference performance on MLPerf v3.0 DLRM benchmarks.

Demonstrates new GPU vendor capability in data-center-critical recommendation inference workloads.
Trade pressSlicast · April 6, 2023 · Global · Source: finanznachrichten.de
importance 65

NEUCHIPS, the leader in AI ASIC platforms for deep learning recommendation, participated in MLPerf v3.0 with their RecAccel N3000 and achieved industry-leading performance and power efficiency. The RecAccel N3000 system delivered 1.7x better perf-per-watt for inference DLRM while maintaining 99.9% accuracy leveraging its INT8 calibrator. These results demonstrate a significant advancement in energy-efficient AI infrastructure for cloud service providers evaluating new data center solutions.

The RecAccel Quad-N3000 PCIe card is expected to increase perf-per-watt 2.2x more than GPU H100 PCIe, offering substantial improvements in performance density. This advancement is expected to deliver the lowest total cost of ownership (TCO) for organizations deploying deep learning recommendation workloads at scale.

MLPerf v3.0 testing was performed on a GIGABYTE G482-Z54 server with an AMD EPYC 7452 32-core CPU containing eight RecAccel-N3000-32G-PCIe cards. During the system testing, the RecAccel N3000 performance resulted in nearly 100% scaling across each card, demonstrating efficient multi-card deployment capabilities.

Kumaran Siva, corporate vice president of Strategic Business Development at AMD, stated: "We deliver high-performance computing products that help intelligent systems effectively manage complex data sets. We are extremely proud of the joint work we have done with NEUCHIPS to achieve leadership performance and power efficiency for DLRM inference in MLPerf v3.0 using AMD EPYC processors. We look forward to continue working with NEUCHIPS to deliver industry-leading AI solutions." Youn-Long Lin, CEO and Chairman of NEUCHIPS, added: "We were thrilled to participate in MLPerf v3.0 and achieve our goal to deliver the world's most energy-efficient DLRM inference platform with the first domain specific architecture. We look forward to working with the cloud ecosystem to support industry sustainability initiatives."

NEUCHIPS Inc., founded in 2019, is an AI ASIC solution provider focused on designing recommendation systems. The company was established by a team of veteran IC and software design experts with decades of experience from leading semiconductor companies and holds patents in signal processing, neural networks, and circuit design.

Read the original
NEUCHIPS' RecAccel N3000 GPU achieved… · Slicast