Ten Year Anniversary of Core 2 Duo and Conroe: Moore’s Law is Dead, Long Live Moore’s Law

Looking To The Future

While today is Conroe’s 10 year anniversary, I was a post-teenage system builder when it was first released. Now, as AnandTech’s CPU editor, it has been fun for me to delve back into the past and revisit some of the fundamental design changes that would steer a significant amount of Intel’s future design. You can certainly feel many of the technologies used in the Core microarchitecture in Skylake today, including operation fusion and large shared caching. Now of course, a number of technologies have been developed since which make a big difference too, such as micro-op caches from Sandy Bridge, an L3 cache, even adaptations for eDRAM and moving the memory controller and north bridge on-die. But it does make me wonder if there will be another Intel microarchitecture as important as this down the line. On the AMD side of the fence, everyone is looking at Zen with wide eyes and anticipation. While we have been told not to expect it to take the performance crown, a number of users and industry analysts hope that it brings more competition to the x86 space, enough to rekindle the competitive spirit in silicon back in the mid-2000s.

Looking into the International Technology Roadmap for Semiconductors report, and even just the 50-page summary, there are a large number of predictions in the industry that could happen. There are thousands of people working to make sure the next process node, and the one after that, happens with good yields and on time. The report goes into detail about how shrinking that process won’t happen forever, which is a sentiment that the industry has had for a while, and it lays out in a series of working groups what needs to happen at each stage of the process to go beyond Moore’s Law, specifically regarding silicon stacking, TSVs, and the movement to 3D chips. The ITRS report is set to be the last, with the new focus on devices, systems, SiP and other technologies beyond Moore’s Law. Some have heralded the lack of a future ITRS report as a stark warning, however the fact that we can’t keep shrinking forever has been a known fact, especially at the point where most businesses won’t shrink a process node unless it can net them an overall profit. The movement to 3D makes everything a lot more complicated, but it has to happen in order to provide semiconductor growth and improvements beyond 2D.

Sources

Johan’s Conroe vs K8 Architecture Deep Dive, 2006
Anand’s Core 2 Extreme and Core 2 Duo Review, 2006
International Technology Roadmap for Semiconductors 2.0 Report, 2015/2016

Addendum: This article originally stated that the Core 2 Duo/Conroe was derived in part from the Pentium Pro. This was due to typo in the original 2006 article and has since been adjusted.