[Hardware] The Promise & Pitfalls of Open Hardware Development

[SliCeR]

The prying open of the hardware stack has begun in earnest, but the precise path this complex process will take has yet to be determined. Market forces, limits of scaling and the emergence of disruptive technologies like artificial intelligence have fueled the rise of open source hardware. Still, as this EE Times Special Project demonstrates, physical, legal and economic barriers remain as a fledgling group of open source advocates and a handful of commercial vendors seek to democratized hardware design. Those proponents and early adopters have focused their energies on reduced instruction set computing, the foundational RISC architecture that emerged from the University of California at Berkeley in the 1980s. RISC has seeded the beginnings of an ecosystem extending beyond processor technology to include open interconnects, network and, ultimately, open computing. As with open source software, key chip makers are eyeing the open hardware movement. Some perhaps with trepidation as semiconductor scaling runs out of steam and monster GPUs and CPUs accelerators approach the end of the line, giving way to new heterogeneous devices and chiplets. Indeed, the nascent open source hardware movement can glean lessons from their software brethren. Over the last several years, in examples of what critics described as “enlightened self-interest,” behemoths such as Microsoft (GitHub) and IBM (Red Hat) have snapped up key sectors of the open source software movement. If and when open source hardware development reaches critical mass, might the same acquisition spree repeat itself, potentially stymying hardware innovation? Industry observers are divided on the impact of pending deals such as Nvidia’s acquisition of chip IP vendor Arm Ltd. would have on the ambitious RISC-V project. “There are concerns expressed that Nvidia’s competitors would be alarmed and maybe move to RISC-V—which Nvidia also has a stake in. But that would be stupid, and Nvidia is not stupid,” said graphics industry analyst Jon Peddie. Regardless of how antitrust and other issues play out, the open computing ecosystem continues to expand. Beyond the RISC-V instruction set architecture (ISA), groups like the Open Compute Project are driving innovations ranging from open interconnects for emerging chiplets to AI hardware-software design collaboration. Indeed, the group’s 2021 strategic plan concludes that a “a co-design approach is needed to allow for simultaneous work on all the [hardware] and [software] bottlenecks which can affect the performance….” The growing complexity of machine learning workloads that dominate enterprise data centers will likely forge closer ties among developers of open code and hardware. We examine these and other issues on our Open Source Special Project. Kevin Krewell, principal analyst at TIRIAS Research, surveys the rise of the RISC architecture and how it underpins much of the open source hardware movement. Indeed, the emergence of the RISC ISA has encouraged open source developers to grab the baton and forge new devices. Analyst Jon Peddie outlines one such project aimed at developing a new, free GPU design dubbed RV64X. Nitan Dahad, EE Times European correspondent and editor-in-chief of embedded.com considers whether the open source hardware movement can duplicate the success of the sprawling Linux ecosystem. He and others conclude the RISC ISA could prove the hardware equivalent of the Linux kernel. As Dahad’s sources explain, however, hardware designers face physical limits that far exceed pushing out the next software release. Contributor John Walko examines how the shift to open hardware such as radio access networks is shaping the deployment of next-generation 5G networks. Meanwhile, EDN Editor-in-Chief Majeed Kamran goes deep into the design details of emerging RISC-V processor designs, concluding that recent progress may foreshadow of new era of open computing. Lastly, we reprise EE Times Editor-in-Chief Junko Yoshida’s interview with Wes McKinney, the brains behind several open source data analytics tools. Among McKinney’s goals is bridging the gap between data science and big data, a chasm that open source hardware can help narrow.