In the last few decades, several times companies and research organizations have proposed potentially disrupting innovations in the semiconductor arena, but almost all the times the Intel steamroller crushed them.
Transputers, innovative RISC architectures, clock-less chips and other technologies that seemed to threaten the existing x86 status-quo on microprocessor design were repeatedly sidelined by or incorporated into classical IC architectures that are still based on the old Von Neumann model, developed by computer pioneer John Von Neumann in the 50s, applied to the x86 computer architecture
So even if now and then a new technology comes out with the promise to disrupt the current chip production ecosystem, I am skeptical and tend to believe that the purported advantages of the technology will be rendered obsolete already by the time when the technology is ready to hit the market.
Timothy Mattson, engineer at Intel’s Microprocessor Technology Laboratory, said once: “If I’ve learned anything in my 20 years at Intel, it is to never bet against our fabs”. In other words, despite being considered one of the most innovation-driven market, microchip production is still very much conservative when it comes to technologies and architectures used.
The reasons are multiple: a natural inertia by the industry that prevents adopting innovative solutions, the issue of back-ward compatibility that prevents large software and hardware companies to mass market revolutionary solutions and the truth that Moore` s Law is driving the industry so fast that new technologies that promise 10x or 100x advancements in a few years lose their appeal once they are ready for high-volume manufacturing.
However, there are exceptions to the rule.
There have been times when a company successfully challenged the incumbent technology and caused a substantial change in the industry: one of such cases has been AMD`s launch of their Athlon processors.
Just when Intel was pushing hard to increase the clock frequency of their processors with the Intel “Tejas” planned to run at 5GHz~10GHz, AMD decided to produce cooler chips that could be used as the base for multi-core platforms.
Finally, “Tejas” got cancelled due to heating problems and Intel never planned to produce a 5GHz microprocessor anymore while AMD used their advantage to gain a substantial amount of market-share and ushered the new era of multi-core processors.
Knowing this little bit of computer history is the reason why I am now cautious about the new announcement that HP is betting a substantial amount of R&D effort on a new technology called “The Machine” and based on a combination of memristors and silicon photonics.
The new “Machine” aims to become a replacement to current computer architectures allowing an unprecedented level of performance.
The logic and memory part of the chip would be both handled by memristors while interconnections would be created using silicon photonics technology, therefore bypassing one of the bottlenecks that prevent performance scaling in contemporary computers.
As some analyst noted, while memristors have been vapourware for many years, they can handle memory and logic using the same platform and therefore eliminating the need of separate RAM and SSDs components.
The use of silicon photonics, moreover, would help create a monolithic system with high-speed communications and no bottlenecks between one part of the system and the other.
From a Business Week article, it has been reported that 75% of HP Labs, the research and development branch of the company, is now focused on “The Machine” project, and this would mean that the company is really betting the farm on it.
While this effort may seem hard to understand or even exaggerated, it must be noted that for years computer makers like HP and IBM have seen their margins progressively eroded by the commoditization of computers (with the likes of Intel and Nvidia grasping all the margins) and competition from Chinese makers like Lenovo.
“The Machine” may then represent the last opportunity for HP to move from a position of underdog in the computer arena and possibly revolutionize an industry that has been based, at its core, on the old x86 architecture, born with the venerable 8086 chip back in the 70s.
If you are interested in our services, please visit our site.
Subscribe to our newsletter to receive our new articles directly in your mail box.
If you liked this article, please give it a quick review in StumbleUpon, Facebook or Pinterest.