We kick off our coverage of Embedded World 2022 this week with AMD’s announcement of a new member of its Ryzen Embedded SoC family. Embedded World runs from June 21–23 this week in Nuremberg, Germany.
Today, AMD has launched its Ryzen Embedded R2000 series of SoCs, which doubles core count, memory bandwidth, and I/O connectivity compared to AMD’s R1000 series processors.
A summary of the specs and features of the new Ryzen Embedded R2000 SoC. This data represents the highest-end model of the family. Image used courtesy of AMD
In this article, we examine the details of the new Ryzen R2000 and how it’s positioned. And we share insights from our interview with David Rosado, Senior Product Manager at AMD’s Adaptive and Embedded Computing Group.
Higher-Performance for Embedded Systems
This new embedded processor news follows AMD’s significant desktop and mobile processor announcements the company made at the COMPUTEX 2022 show in late May. But, appropriate to the Embedded World venue, this new Ryzen Embedded R2000 family of devices is aimed at the embedded market, and in particular industrial systems. According to AMD, the new family is tailored for industrial and robotics systems, industrial IoT, vision systems, and thin-client systems.
Graphical comparison of the four Ryzen Embedded R2000 family SoCs. Image (modified) used courtesy of AMD
The Ryzen Embedded R2000 Series doubles the core count of the “R” series from two to four cores and up to eight threads, with clock speeds up to 3.35 GHz. Based on AMD’s “Zen+” x86 core architecture, the devices also embed up to eight Radeon GPUs (graphics processing units). The devices offer up to 2 MB of L2 cache and 4 MB of shared L3 cache.
For memory, the high-end R2544 family member provides two channels of 3,000 MT/s DDR4 DRAM with ECC support. This is a jump from the 2,400 MT/s DRAM of the high-end R1000 family device, the R1606G.
In line with the system-on-chip (SoC) nomenclature, the R2000 devices provide a rich set of high-speed peripherals and interfaces. Beyond the 16 lanes of PCIe Gen3, the chips provide two SATA 3.0 ports and six USB ports (USB 3.2 Gen2 and 2.0). Operating system support includes Microsoft Windows 11 and 10, along with Linux Ubuntu LTS.
Performance Uplift from First Gen R1000
AMD is calling the R2000 family a “generational performance uplift” from its earlier R1000 line. Where the R1000 processors supported up to three 4K display interfaces, the R2000 chips can link to up to four independent 4K displays. The devices work with DisplayPort 1.4, HDMI 2.0b, or eDP 1.3 interfaces. The R1000 family supported up to eight PCI Express (PCIe) Gen 3 lanes, while the new R2000 family accommodates up to 16 Gen 3 PCIe lanes.
The Ryzen Embedded R2000 family includes four models, two of which are now in production. Image used courtesy of AMD
Importantly, the new R2000 processors share the same FP5 BGA package and footprint as the previous R1000, meaning system developers can upgrade to the new devices without redesigning their PCBs. As it does for all its Ryzen Embedded devices, AMD ensures planned availability for the R2000 SoCs up to 10 years, providing system designers a long-lifecycle roadmap.
Like its R1000 predecessor, the R2000 family supports the AMD Secure Processor to help guard sensitive data and validate code prior to executions. Support is also provided for AMD Memory Guard, a suite of security features comprising Secure Boot and Secure Memory Encryption—basically real-time DRAM encryption. Using those technologies, stored memory is encrypted to keep an attacker from accessing the data. This helps to prevent cold boot attacks.
Three Performance Benchmarks
In order to illustrate the increase in performance between the Ryzen R1000 series and the new Ryzen R2000, AMD shared benchmark information comparing the two. Rosado explains each of the benchmarks. Cinebench R20 is a popular free benchmarking tool that’s good at measuring hardware system capability, says Rosado. It’s a benchmark that puts pressure on all the processor cores at once using a unique rendering scheme.
A second benchmark AMD used is 3D Mark 11. Rosado says It involves measuring the movement of 3D pixels and is a good indicator of graphics performance modeling, 3D models, and so forth. The third benchmark is PassMark 10, which Rosado describes as a scheme that measures the CPU as a unit—how much data a CPU can crunch independent of any interfaces that could bottleneck performance.
The new Ryzen Embedded R2514 offers an 81% increase in performance over the first-gen R1606G SoC. Image used courtesy of AMD
As the chart above shows, AMD used the first generation Ryzen Embedded R1606G as a baseline (100%). The most dramatic results were in the new Ryzen Embedded R2514, showing a 68% improvement in a multithread (n) Cinebench comparison, an 81% improvement in 3D Mark 11, and an 81% leap in the PassMark 10 benchmark.
More Data, More Data Processing
The R2000 was designed to serve several markets, but industrial IoT exemplifies the problem that the SoCs solve. According to Rosado, the primary need for the R2000 family of SoCs is all about dealing with increased amounts of data processing needed at the industrial IoT edge.
“When an industrial machine is making a gazillion widgets in a day, you’ve got a ton of data. What was the position in the conveyor lane line? Does the widget have any imperfections? There’s a ton of data, and you have to decide, what goes to be processed in the cloud versus what doesn’t.”
Higher performance embedded processing, such as the R2000 provides, enables that processing to be done locally. “Today, even machine learning at the edge is becoming very relevant because you want to generate an inference model locally instead of having to generate it on the cloud,” says Rosado.
AMD will be showcasing the Ryzen Embedded R2000 family this week at Embedded World, along with several other demonstrations, in Hall 3A, Stand 239 at the show. A product brief is available summarizing the R2000 family.