The AMD Embedded R-Series platform further supports up to four independent displays, or one extremely high-resolution display with a resolution of up to 4000x2000 pixels. This is particularly relevant for large digital signage screens that need to deliver sharp images even at close range. Another special feature that it includes is AMD Dual Graphics technology, which combines the graphics performance of the integrated GPU with a discrete graphics unit, such as the AMD Radeon E6760 GPU. This means that when the discrete graphics unit is used, the overall graphical performance of the system is improved, whereas previously, it was only possible to use either one or the other graphics unit. For the first time, developers can now efficiently scale the graphics performance of their embedded application without incurring power loss or extra costs. Thanks to AMD Eyefinity technology, it is also possible to control up to 10 independent displays via a single APU and discrete AMD Radeon graphics processor. This opens up new levels of efficiency and performance for infotainment and digital signage installations. For example, congatec has already developed a nine-screen digital signage system based on the new AMD Embedded R-Series platform that demonstrates the impressive capability for large-scale video wall installations – see figure 2. With the new AMD R-Series APU, developers can enter directly into the application development. The APUs support all current APIs of DirectX 11 including DirectCompute plus OpenGL 4.2 and OpenCL 1.2, which can speed up application development significantly.
The COM concept
One of the best ways to incorporate these new APUs in an application is by using Computer-on-Modules (COMs). They provide developers with the comfort and safety of a classic board solution combined with the flexibility of custom designs. COM manufacturers provide compatible, application-ready modules that are pre-integrated for many standard and real-time operating systems and also include all necessary peripheral drivers.
Thanks to close cooperation between module suppliers and chip manufacturers such as AMD, developers are able to integrate the latest technologies into their own designs shortly after the new building blocks become available, without complicated and expensive custom development. This cuts development time and costs quite dramatically.
Even though the modules themselves are standardized, a lot of useful know-how lies in the details, such as the scope and quality of the supplied firmware and software or operating system modifications. Important system components can be either pre-integrated on the COM already or can be added as an optional configuration. This is especially important for expanding the battery management or when adding a graphics unit for designs that provide extreme scalability of processor and graphics performance.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.