AMD announces Ryzen Embedded 8000 series processors for ‘Industrial AI’

April 3, 2024

AMD has recently announced a ‘breakthrough innovation’ in the realm of industrial Artificial Intelligence (AI) with the introduction of the Ryzen 8000 series embedded processor. According to AMD, these processors promise a significant leap forward in handling demanding workloads within AI applications, setting a new standard for processing performance, integration, and onboard AI acceleration. In this article, let’s dive into discussing the AMD Ryzen Embedded 8000 Series in more detail.

AMD Ryzen Embedded 8000 Series: Overview

Embedded Processors

The Ryzen 8000 series processors are the first to integrate a Neural Processing Unit (NPU) built on AMD’s innovative XDNA architecture. This integration changes the landscape of industrial computing by enabling targeted, efficient AI acceleration for improved computational agility and reduced power consumption. AMD states that NPUs bring forth a new era of AI processing by providing specialized inferencing at the processor layer, catering to applications such as machine vision, robotics, monitoring, and human-machine interfaces (HMI).

These processors deliver a range of thermal design power (TDP) profiles, spanning from 15W to 54W. With a top maximum performance of 16 trillion operations per second (TOPS) delivered by the NPU, and the full processor performance scales up to 39 TOPS.

New Chipsets

Some months prior, AMD unveiled the 8000 series. However, in this latest release, they introduced two new processors in the embedded series: the 8845HS and 8645HS. Both chips boast 16MB of L3 cache and a 45W TDP, with maximum turbo frequencies reaching up to 5.1GHz and 5.0GHz respectively. Additionally, the 8845HS offers an 8-core/16-thread configuration, while the 8645HS features 6 cores and 12 threads.

Connectivity and Others

The Ryzen 8000 series processors ensure efficient data transfer with 20 lanes of high-speed PCIe Gen 4 connectivity. Moreover, support for dual-channel DDR5 5600 memory with error correction (ECC) enhances reliability and keeps operations running smoothly. These processors can also power four independent 4K displays, enabling visually stunning output for industrial applications.

With 20 lanes of on-chip PCIe Gen4, ample I/O flexibility is provided for specialized accelerators and other add-ons. The inclusion of dual-channel memory support and the latest AMD RDNA 3 graphics technology optimizes visual processing and multimedia performance, catering to the demands of industrial systems requiring graphical consoles and status panels.

Also, read

AMD Ryzen Embedded 8000 series Specifications

CPU Architecture: “Zen 4”
Process Technology: TSMC 4nm
CPU Cores/Threads: Up to 8 Cores, 16 Threads
L1 Cache: L1 Instruction Cache = 32 KB, L1 Data Cache = 32 KB (per core)
L2 Cache: L2 Cache = Up to 8 MB (total)
L3 Cache: L3 Cache = Up to 16 MB unified
Graphics: RDNA 3 graphics with up to 6 WGPs (Work Group processors)
Memory: 128b DDR5-5600 with ECC
NPU: 16 TOPS
Security Processor: Microsoft Security processor to provide fTPM 2.0 and additional security features
PCIe: x20 PCIe® Gen4
USB Support:
- 2x USB4 Type-C with DP Alt-mode,
- 1x USB3.2 Gen2 10Gb/s Type-C DP Alt-mode
- 1x USB3.2 Gen2 10Gb/s
- 3x USB2.0
- 1x USB2.0 (SecureBio dedicated)
Max Turbo Frequency: Up to 5.1GHz

AMD Ryzen Embedded 8000 Series Availability

While the Ryzen embedded 8000 series initially targets industrial AI applications, it is anticipated to become available for general consumers sooner rather than later.