It’s been an interesting couple of weeks over at Intel. I had meant to do a write-up on all this breaking info weeks ago, but it kind of got away from me. Consider this my catch-up post. Intel debuted some interesting new technology a few weeks ago that will radically change the performance and energy efficiency of its products. As well, the company has announced some changes to their roadmap that is indicative of the changing PC market and Intel’s place in it.
Starting off, we have Intel’s move into the next dimension with 3D transistors. No, these have nothing to do with 3D images or televisions. 3D transistors – what Intel calls their “tri-gate” – work by changing the physical shape of individual transistors in order to allow more voltage through each “gate” – the on/off switch that controls the flow of electrons and allows for binary addition that makes up processors. Traditional transistors are planar and have one surface of silicon for electrons to flow on. In order to increase the performance of transistors, you need to either increase the voltage through the transistor or make it smaller so that the same amount of voltage travels shorter distances. The problem is that as transistors have gotten so small, this strip of silicon is becoming only 100 or so atoms wide. The gate that decides whether a transistor is “on” or “off” has a harder time doing that job on such a tiny scale. Plus, as transistors get so tiny and more and more are added to processors, this inability to judge whether gates are on or off causes leakage, adding heat and power drain to chips.
The phenomenon of “leakage” has been the bane of Intel for over a decade. As transistors shrunk below 90nm (that’s 90 billionths of a meter in size) this became a serious problem. Normally smaller chips should be more efficient and faster, but only if you can combat leakage. So in order to combat this as Moore’s Law demands smaller and faster transistors, Intel and other chip builders have turned to insulator technologies. This applies a layer of insulating metal and oxides on chips which prevent leakage. SOI (silicon-on-insulator) was a popular choice for the 90nm and 65nm nodes from companies like AMD and IBM. Intel started using hafnium – a rare element – as an insulator for its 45nm and 32nm products. However, on the eve of the 22nm revolution, normal insulator technologies are becoming insufficient. Intel’s tri-gate is a radical change that alters the very shape of transistors. Now, instead of this single plane of silicon, it is shaped like a fin. This provides three surfaces for electrons to flow through, thus allowing better conductivity and much faster performance. Intel states that this increases performance by 37% and reduces power draw by 50%. Tri-gates will debut with Ivy Bridge, the 22nm shrink of Intel’s current Sandy Bridge processors. While not architecturally different, the use of Tri-gates could mean major performance and efficiency improvements the likes of which we have never seen in a simple die shrink.
This increased performance with enhanced efficiency bodes well for Intel’s revamp of their notebook processors. For years, Intel’s sweet spot for TDP on notebook chips was 35W. Extreme chips were a bit higher; low and ultra-low voltage are obviously lower in their power draw. But 35W was the standard. Intel sees this as insufficient for the next generation of notebooks that need to be thin, light, and last all day on a single charge. In essence, notebooks are moving towards the Samsung Series 9 or Macbook Air. Intel’s new standard will be 15W, effectively making ULV chips the new standard and 35W where more performance-oriented products will go. Before the Tri-gate, this seemed pretty much impossible. Sandy Bridge ULV processors are not slouches, but the meat and potatoes of the line comes in their quad-core offerings. Tri-gates will allow more performance to be squeezed out of these tiny chips than ever thought possible. It’s a great move for Intel, which needs to change with the times and offer more flexibility in smaller form factors. The company is also accelerating their Atom roadmap and plans to debut processors with tri-gates by early 2012, finally getting these processors up to snuff in the tablet and mobile phone areas.
File this one under “interesting.” Intel has made some interesting comments this morning in a Reuters interview which basically asserts that Apple, one of Intel’s highest profile clients (yet one of the smallest in terms of products sold), is actually quite integral to Intel’s chip roadmap. This revelation is an interesting one because it breaks the notion that Intel and Microsoft are as tightly knit as once thought. Perhaps the reign of “Wintel” has come to an end?
Intel’s Atom processor has seen a tumultuous but brief history in the CPU space. Designated as Intel’s super efficient and extremely small chip solution, many figured Atom would be Intel’s way into the tablet and smartphone space. But after 4 years Atom remains out of that space, completely dominated by ARM. Instead, Atom got its lease on life in the netbook market – a briefly popular segment that filled many computing needs during the recession.
It’s called Sandy Bridge. We’ve known it existed for a few years and would debut after Nehalem in the 2010-2011 time frame. Today however, we now have a better idea of what exactly we can expect from the new architecture. In short; it’s more evolutionary than revolutionary, and it looks to be fixing many of Nehalem/Westmere’s minor woes.
A few months back I did a two-part series called “Upheaval in the GPU Wars” which chronicled the ups and downs of Nvidia and ATI over the decade in their race to offer the best graphics solutions to consumers. As I mentioned in the last post, I had intended for that series to become a three-part affair as I would address Intel and their history of offering sub-par graphics solutions and its effect on the industry. However, the format I was using was rather long-winded to get to the point, and chronicling Intel’s history in this area would be effortlessly boring charting failure after failure. To top that off, it wouldn’t really get to the point of talking about Intel’s problem. As such, I’m going to frame the discussion of Intel’s graphics issues in a different light, directly addressing the problem of their horrid performance.
Intel’s Core microarchitecure in the Core 2 line was a major win for Intel after years of falling behind to AMD. These processors were extremely successful at offering great performance, good power management, and nice scalability in clockspeed and core count. However, after 3 years or so, Core 2 has grown long in the tooth and Intel has since pushed it down into the budget segment. Replacing these venerable processors has been Intel’s Core i-series, which is a blanket marketing name for any processor built under Intel’s “Nehalem” microarchitecture. I’ve discussed the advances in Nehalem over Core before but to summarize, Nehalem brings the following new features: monolithic (single die) quad/hexa/octo-core processors; Dynamic Turbo Boost, which allows cores to shut down and speed up active cores; Hyperthreading, which allows one processor core to handle the work flow of two cores; L3 cache, which gives the processor more onboard memory to access for accuracy and less access to RAM; integrated memory controllers, which makes accessing memory much faster; and a new bus system, which connects the processor to other system services at a much faster pace. All of these technologies are meant to improve overall system performance and maximize data access over Core 2, which had a problem of often being starved for data because of system bottlenecks. Overall, Nehalem has become a powerhouse chip architecture that has no rival in the consumer space.
Digital Gravitas 