This article was originally published on May 10, 2021
Editor’s Note: This content is contributed by Chetan Khona, senior director of Industrial, Vision, Healthcare & Sciences, AMD.
Big ideas don’t usually require a lot of words. It’s been my experience, the more words you use, the more specific, and smaller, an idea becomes. When we started developing Kria™ system-on-modules (SOMs), we rallied behind the audacious concept of, “AMD benefits without FPGA design.” What a rabbit hole that led us down!
We took a holistic view of the entire FPGA design process and stacked a series of fairly ordinary and unassuming concepts on top of each other to build something unique and multi-faceted. For anyone out there that has ever said, “Why am I spending so much time on printed circuit board (PCB) trace length matching for this DDR memory interface?” Or perhaps, “I need how many power supplies and in which order do they need to ramp?” We took both concerns off the table with the production SOM concept. Or how about those that have uttered, “But I’ve never done an FPGA design!” No worries, we took care of that as well with a growing app store full of pre-built designs, that could be, but don’t have to be, modified. Many more large and small innovations exist and are expanded upon in a white paper titled, “Achieving Embedded Design Simplicity with Kria SOMs.”
The one variable that has been fixed to enable the simplicity of the Kria experience, was the scope of applications. Rather than attempting to “boil the ocean”, with the KV260 vision AI starter kit, the focus is uniquely smart vision applications. Expect more starter kits expanding the scope of applications in the future, while holding true to the Kria concept of software-controlled, adaptive hardware acceleration to meet shifting requirements presented to the user. In the meantime, after you read the white paper, check out the Kria page on our website and register for the on-demand webinar, to help you evaluate, design, deploy, and adapt with Kria SOMs.
