The big news from Intel’s Computex 2017 keynote yesterday was the new performance line of Core X processors, but they did manage to slip in a little info about the upcoming refresh of their good old Core processors. The 8th generation Coffee Lake CPUs are in development now, and during the press conference, Intel VP Gregory Bryant confirmed plans to ship them around the holiday season.
Coffee Lake, like 7th generation Kaby Lake, does not follow the usual tick-tock pattern of Intel releases. For a while, Intel was able to shrink their transistors once every two years, allowing them to create higher-density and more powerful chips. That cadence is referred to as Moore’s Law, but Moore’s Law has had to undergo a revision — since shrinking their transistor die to 14 nm with their 5th generation chips, Intel hasn’t moved on.
That means Coffee Lake will be year four of 14 nm. But, that doesn’t mean Intel isn’t making progress. The company talked transistors at their Tech and Manufacturing Day in San Francisco in March, and without getting too deep into the nitty gritty, Intel explained how they’ve been able to increase transistor density using those 14 nm transistors year over year. That’s allowed them to wring out more performance with each generation, even if we’re not seeing the huge leaps forward we used to see.
An interesting wrinkle to watch for is the PCH in use. We’ve been on X99 since the 5th generation of Core chips, too, and with the modest gains in performance expected this year, it’s likely this won’t change. A new PCH, X299, will be introduced with the Core X line, but it seems unlikely that Coffee Lake will follow suit (although this is unconfirmed). This is important for those who build their own computers — no change in PCH means that it’s not necessary to buy a new motherboard to support the latest processor.
This year, Intel says to look for performance gains of up to 30 percent. But, there’s a caveat — they compared a 7th generation i7-7500U to estimates for its 8th generation counterpart. Not only are we dealing with estimates and not benchmarks, it’s also worth noting that the 8th generation i7 U-series CPU is slated to have four cores and eight threads, instead of the two cores and four threads found on the 7th generation chip. Long story short, huge gains between 7th and 8th generation chips are unlikely — if you want a big jump up in power, you’ll want to look to those new Core X processors.
Otherwise, we’ll have to keep wondering when that next long-awaited die shrink will happen. Intel ditched the tick-tock schedule at their Tech and Manufacturing Day, replacing it with process-architecture-optimize — a cycle that involves shrinking the transistors, redesigning the chip to fit more of those transistors, and then finding ways to squeeze as much power out of that configuration as possible. It’s not meant to be a hard three-year cycle, either — Intel will optimize as often as they want until they feel it makes sense to shrink the transistors again.
But, they will shrink them eventually. Intel has plunked down $10 billion on a fabricator under construction in Arizona for the purpose of churning out 7 nm chips. You can bet that once it’s fully operational (in 2019 or 2020), Intel won’t just let it sit there — especially with rival AMD looking game for the first time in years.
Check out more of our Computex 2017 coverage right here!
Disclosure: Taiwan External Trade Development Council (TAITRA), an organizer of COMPUTEX, covered our travel expenses to attend COMPUTEX 2017. All thoughts and opinions are 100% our own.