Intel's range of sixth-generation Core processors, codenamed Skylake, is now public. And boy, am I disappointed.
Most of the fifth-generation Core processors have successors, at least approximately. The Broadwell generation's rollout was slow and messy, with Intel apparently struggling to get its 14nm manufacturing process as refined and as reliable as it wanted. The announced Skylake lineup is more complete, and it shouldn't take so long to come to market, so that's an improvement. But amid all that Broadwell mess was a truly monstrous chip: an almost mythical beast, the Core i7 5775C. What made this part so special? It paired two features. One is mundane: the processor is socketed rather than soldered, meaning that enthusiasts can use it in self-built systems, pairing it with the precise range of components that they want.
The other is rather more exotic: the chip has Iris Pro graphics, and with it, 128MB of eDRAM. Crack open the processor and it has not one big chip in its package but two, with the eDRAM nestled alongside the processor itself. The RAM is primarily there to accelerate graphics operations, but Intel's design means that it is not dedicated to this task. For Broadwell, it functions as a large, high bandwidth level 4 cache (the other 3 levels being part of the processor chip itself). Skylake shakes up the design somewhat, changing the topology and allowing the eDRAM to cache even more stuff, but the effect is still the same: a monstrously large cache for a mainstream commodity processor.
And if Broadwell is anything to go by, that cache does real work. Tech Report's review of the first Skylake processors included scores from the 5775C, and in games the performance was remarkable. The 65W 3.3-3.7GHz i7-5775C beat the 91W 4-4.2GHz Skylake i7-6700K. The Skylake processor has a higher clock speed, it has a higher power budget, and its improved core means that it executes more instructions per cycle, but that enormous L4 cache meant that the Broadwell could offset its disadvantages and then some. In CPU-bound games such as Project Cars and Civilization: Beyond Earth, the older chip managed to pull ahead of its newer successor.
The effect was far from universal. The 5775C gives up a lot in clock speed (and power consumption) to the 6700K, and with that advantage, the Skylake part often wins. But in memory-intensive workloads, such as some games and scientific applications, the cache is better than 21 percent more clock speed and 40 percent more power. That's the kind of gain that doesn't come along very often in our dismal post-Moore's law world.
Those 5775C results tantalized us with the prospect of a comparable Skylake part. Pair that ginormous cache with Intel's latest-and-greatest core and raise the speed limit on the clock speed by giving it a 90-odd W power envelope, and one can't help but imagine that the result would be a fine processor for gaming and workstations alike.
But imagine is all we can do because Intel isn't releasing such a chip. There won't be socketed, desktop-oriented eDRAM parts because, well, who knows why. The processors announced so far do not include a socketed eDRAM part. Intel still has more processors to announce—including various Iris and Iris Pro parts that will include eDRAM—but none of these will be socketed desktop parts. At Intel's Skylake briefing last month, Tech Report's Scott Wasson asked if such a chip would be released and was explicitly told that, like "fetch," it's not going to happen.
It's true that, like the 5775C, this would have been a niche processor with a price premium, but that's not really something that has stopped the company before. Intel has made many processors aimed at the performance-hungry enthusiast segment; the entire purpose of the overclockable K parts is to appeal to those customers.
More interestingly, perhaps, Intel appears to be willing to make changes to its processor designs that are at least partially motivated by the demands of overclockers. In previous Intel processors, the chips used the same base clock speed for both the processor's integrated PCIe controller and everything else. This meant that overclocking by increasing the base clock speed was undesirable, as it also overclocked the PCIe bus. In Skylake, those clocks are separated from one another, enabling base clock to be increased without pushing the PCIe bus out of spec. This makes overclocking more flexible and reliable—it opens the doors to overclocking both by increasing the multiplier and by increasing the base clock, instead of just the multiplier—but doesn't immediately appear to have any other big upside. In comparison to this change, shipping socketed eDRAM parts surely wouldn't be much extra work at all.
I want to put together a new PC, and I want it to be as fast as is practical without venturing into the exotica of thousand-dollar processors. I also want it to use Intel's latest-and-greatest architecture because there are new chip features that I want to explore. But my PC is also a working tool, and as such, I'm unwilling to overclock it. The risk of additional instability may be low, but it's non-zero. But a chip that gives me a bunch of extra performance just by including a cache that's bigger than the first hard disk I ever owned? I'd be glad to throw some money Intel's way for that.
Intel could have had a Skylake processor that was exciting to gamers and anyone else with performance-critical workloads. For the right task, that extra memory can do the work of a 20 percent overclock, without running anything out of spec. It would have been the must-have part for enthusiasts everywhere. And I'm tremendously disappointed that the company isn't going to make it.
reader comments
146