- In a ceremony at The Computer History Museum in Mountain View, California, AI company Cerebras Systems was honored for continuing this legacy.
- The WSE-2 chip and its predecessor constitute a milestone in merging transistors, a fundamental electronic element, into a single component.
In today’s world, technology is one of the most preservative practices, meaning that every invention builds upon the successes and failures that have come before it.
In a ceremony at The Computer History Museum in Mountain View, California, Cerebras Systems, a computer systems company dedicated to accelerating deep learning, was honored for continuing this legacy. The Museum has installed an exhibit featuring the “Wafer-Scale Engine 2,” or WSE-2, the second edition of the company’s AI processor that is the world’s largest computer chip. Cerebras released the chip a year ago to power new versions of Cerebras’s supercomputer, the CS-2.
Andrew Feldman, co-founder and CEO of Cerebras, said, “It is the honor of a lifetime to be inducted into the Computer History Museum’s world-renowned collection.”
The emergence of Cerebras is remarkable not just because of its technological achievements but also because of its ramifications for humanity.
Dan’l Lewin, President and CEO of the Computer History Museum, said, “Being able to optimize compute in the manner Cerebras’s chip does, it has implications that are “stunning” for humanity. It will hopefully teach us that there are more potential positive uses of these technologies than the unfortunate things that have occurred due to some business models that have surfaced that are, effectively, programming the population without them being aware of it.”
The WSE-2 chip has 2.6 trillion transistors, about fifty times as many as Nvidia’s largest GPU chip to date, in a silicon substrate of 46 square millimeters, nearly the entirety of a twelve-inch semiconductor wafer from which several chips usually are sliced.
The integrated circuit features 850,000 “cores” to perform AI instructions in parallel.
WSE technology created a single chip that uses an entire wafer, a chip industry goal for decades. Cerebras’ success came in part from learning from earlier mistakes.
Andrew Feldman, co-founder and CEO of Cerebras, said, “It means a lot to us that this institution has recognized the scope of the effort that had crashed and burned previously in other incarnations. Even by some of the founding fathers of our industry — even Gene Amdahl couldn’t get it to work.”
Gene Amdahl, a mainframe computer pioneer, tried but failed to build a monolithic mainframe component in the 1980s. His attempt led the chip industry to believe that making a wafer-sized single chip was practically impossible.
When Cerebras approached the challenge 30 years later, the wafer-scale endeavor was possible because of long-standing improvements in silicon manufacturing techniques and chip design software tools.
Dan’l Lewin said, “The Cerebras breakthrough has in one sense to do with broadening people’s access to digital tools that were previously the captive realm of specialists.”
Lewin further said, “Industries go through transitions from vertical to horizontal. There used to be these huge industries that were very vertical that were targeted at automating rational tasks like the word processing industry, and CAD/CAM as an industry,” he said, referring to computer-aided design. The progress of microprocessors led to horizontal apps such as Microsoft Office that made those previously vertical industries “approachable by so many people.”
Lewin also stated, “The Museum’s award is a peek into the future as much as a rumination on the past. “History is not about the past; it’s about the present having a conversation with the past.”
Grady Booch, a programming pioneer, was honored at an award ceremony for his several innovations, including Unified Modeling Language.
Dan’l Lewin shared, “And so, this handshake and accelerated opportunity, by taking this immense capability and optimizing it in this way, will help drive these changes, these shifts from horizontal to vertical — they’re being compressed in time.”
Similar to getting a lifetime achievement award, the distinction of being shown in a museum may feel like the end of the world.
Andrew Feldman said, “Museums are often warehouses of the past. We are now building a great company on the back of some pathbreaking technology, and we don’t forget that: both are hard.”