Techday
This page was machine-translated and may differ from the original. View original

AMD Expands Ryzen AI Embedded P100, Accelerating Its Conquest of Industrial and Edge AI Markets

Google 우선 소스Published2026.03.10 14:42

Zen 5 CPU, RDNA 3.5 graphics, and XDNA 2 NPU integrated on a single chip
80 TOPS AI acceleration performance, low latency, and power efficiency

As AI technology rapidly spreads across industrial settings, including factory automation and autonomous robots, demand is growing for edge computing platforms that simultaneously satisfy real-time processing and stability.

AMD announced on the 10th that it is expanding its Ryzen AI Embedded P100 series processor lineup and is targeting the industrial and AI edge markets.

The newly added P100 series maintains the same compact BGA package as existing products, while doubling the number of CPU cores and improving GPU computing performance by up to eight times.

The overall AI computational performance of the system has also been improved by approximately 36%, designed to handle complex edge AI workloads more efficiently.

This processor integrates 8 to 12 Zen 5 CPU cores, RDNA 3.5-based graphics, and XDNA 2 architecture NPU into a single chip.

This provides AI acceleration performance of up to 80 TOPS, while simultaneously securing low latency and power efficiency.

It is optimized for industrial environments with an industrial temperature range, 24-hour continuous operation, and a long-term supply policy of up to 10 years.

Especially the AMD ROCm software ecosystem With this added support, developers can easily deploy AI models to edge environments by leveraging existing open-source AI frameworks.

The HIP-based architecture, which enhances the portability of GPU programming, reduces vendor lock-in and enables scalable system design.

The P100 series is expected to be utilized in various fields, such as machine vision in industrial PCs, spatial recognition and path planning in autonomous mobile robots, and 3D image processing in medical imaging equipment.

Compared to the previous generation, multi-threaded performance is improved by up to 39%, and the overall AI processing performance of the system is more than doubled, providing stable performance even in large-scale language models and virtualized environments.

AMD also provides a virtualization reference stack based on the Zen hypervisor, enabling independent operation of Linux, Windows, and RTOS on a single platform.

This enables predictable real-time performance even in industrial mixed-criticality applications.

Currently, global ODM companies such as Advantech, Congatec, and Kontron are introducing commercial solutions based on P100, and samples of 8-12 core products are being supplied.

Mass production is scheduled to begin sequentially in the second half of 2026.
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.
배종인 기자
배종인 기자