![]() ![]() The vision will play out over time – maybe a decade or more – but it will play out. "Everything I've described here today is happening in labs across the world. "I'd like to remind everyone that these are not just cartoons on paper," said Koduri. Combined, they could increase the number of transistors being placed on computer chips almost 50-fold compared to today. ![]() Thats why its such big news that a team at Lawrence Berkeley National. This would provide yet another doubling.Īll of these five steps are shown in the animation below. The smaller your transistors, the more you can fit on a chip, and the faster and more efficient your processor can be. The final step, Koduri predicted, could be die-to-wafer stacking. So the next step may involve wafer-to-wafer stacking, enabling a further 2x boost in transistor count, or a cumulative 24-fold increase. Instead, die stacking and packaging technologies would provide the increased density. Regular pitch scaling would no longer be feasible beyond this point. These technologies are already being researched, he added. NMOS and PMOS nanowires could then be stacked on top of each other, for another doubling, or a cumulative 12-fold increase. The FinFET architecture would then morph into nanowire architecture, he explained, doubling the previous transistor count for a combined six-fold increase. ![]() The first step would focus on current FinFET transistor technology (10nm), with further improvements in pitch scaling, tripling its density. Koduri outlined several key advances that could shape the next 10 years of computer processor development. "We firmly believe there is a lot more transistor density scaling to come," said Koduri. Koduri's prediction is right, the first chips with over a trillion transistors may emerge by the end of this decade (although, strictly speaking, Cerebras Systems achieved that particular milestone last year their chip is, however, spread over a much larger area than conventional form factors). Top-of-the-range processors currently hold about 30 to 40 billion transistors. Koduri is adamant that plenty of life remains in this trend. Depending on the doubling time used in the calculations, this could mean up to 100 fold increase in transistor counts on a chip in a decade. However, thanks to new innovations in processor architecture and power consumption, Mr. Intel's Chief Architect, Raja Koduri, has presented a roadmap for increasing the number of transistors able to fit on a chip by a factor of 50.ĭuring a keynote presentation at this year's Hot Chips conference (held virtually), he described the ways in which computer technology can continue to shrink over the next 10 years – helping to sustain the famous trend known as Moore's Law.įor many years, analysts have been predicting the end of Moore's Law, with concerns that the exponential growth in computer processing power may be slowing or about to reach a fundamental limit. 50-fold increase in transistor density is possible by 2030 ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |