Today, Intel announced the imminent arrival of its newest stack of desktop processors—the 12th Gen Intel Core series, codenamed “Alder Lake” during development—comprising six fresh additions to its consumer-oriented stable of Core i9-, Core i7-, and Core i5-branded CPUs. Debuting on Intel’s “7 Process” (read more at the link about why that might not mean what you think it does), the chips will be going on sale on November 4, starting at $264 for the Intel Core i5-12600KF. Most important: They will be the first Intel desktop chips to deploy an ARM-style “big.LITTLE” design of mixed core types for power and power savings. (Much more on that in a moment.)
In a time when AMD’s mainstream consumer Ryzen chips are hard to find at their MSRPs, and many of the 11th Generation Intel chips (“Rocket Lake,” now clearly a dead-end platform) are gathering dust on shelves, Intel is in a prime position to offer its latest lineup at an attractive price point. But what kind of upgrades and improvements can you expect out of what Intel is calling the world’s best gaming processor, and will they be worth the early-adopter fee of buying into a whole new platform? Let’s dig into the news to see what’s in store for Intel’s latest launch.
‘Alder Lake’ Splashes Out: 12th Gen Core Series Specs and Pricing
For starters, let’s jump right into the CPU stack and what we learned about it last week during an Intel pre-brief for the tech press. In comparison to last year’s launch of 11th Generation Core (“Rocket Lake”) with 14 separate chips up and down the list, this time Intel has trimmed things down considerably, to just six processors in the initial wave…
Really, it’s just three different chips; the models ending in “F” are the now-traditional alternate versions without on-chip graphics. The first thing you should notice right off the bat in this spec sheet is the unusual relationship between the number of cores on a chip and the thread count. While traditionally Intel, at least for the past two generations, has Hyper-Threaded every core on the processor (one core supports two processing threads), this time around only a subset of cores per CPU have this support.
This is down to the “big.LITTLE” chip architecture style that Intel has opted for this time around, which the company is referring to under the umbrella term “Intel Performance Hybrid Architecture.” In this design, two different types of cores are paired together on the same die. In the case of Intel’s 12th Generation desktop chips (and likely for the coming laptop equivalents, but no news on that until at least 2022), these take the form of what Intel is referring to as “P-Cores” (“P” being short for “performance”), and “E-Cores” (with the “E” representing “efficiency”).
For example, the Intel Core i9-12900K has “16 cores,” however only eight are P-Cores—the same count we saw in this year’s 11th Gen Intel Core i9-11900K—while the other eight are E-Cores. This means you should expect roughly “eight cores’ worth” of gaming performance, minus the impact of background apps, while content creation tasks should see a considerable boost in performance over last generation with the added core and thread count.
During our press briefing, Intel was keen to focus much of its marketing toward the gamer/streamer market, though a few content creation benchmarks made an appearance too.
The point of this approach? Use clever scheduling, via a new Windows 11-only feature known as Intel Thread Director, to move lightly threaded background apps—think Discord, Spotify, Slack—over to the E-Cores, while any heavy single-core loads—games, video editing/rendering, content creation—are handled by the P-Cores.
Intel worked closely with Microsoft to develop its new scheduler to make sure that Thread Director plays nicely with Windows 11, and according to the company, the result is a more harmonious relationship between software and hardware (perhaps akin to the kind of pinpoint-accurate scheduling we’ve seen with Apple, macOS, and its M1 chips).
So, the reason we see that non-linear relationship between threads and cores is because this time around Intel is only Hyper-Threading the P-Cores. E-Cores will support just one processing thread per core. How those threads are divided, moment to moment, is up to Thread Director, and only Windows 11 users will be able to take full advantage of everything the 12th Generation chips have to offer.
Why P-Cores and E-Cores? Intel says it went in this direction to keep power draw and heat in check while the processor is working at its hardest, while also adapting to new use cases, such as streaming your gameplay to sites like Twitch. In a scenario like this, the game would be handled by the P-Cores, while the applications necessary to run a stream—OBS, chat tools, donation dashboards, and so on—are all sidelined to the E-Cores.
This should, in theory, remove the need for streamers to buy two separate computers—one for gaming, one for processing the stream upload—to keep their bitrate at high enough quality levels. However, remember: All of this is predicated on the idea that you’ll be using 12th Gen to stream gameplay on an updated, licensed Windows 11 machine—Thread Director will not be available on other versions of Windows, or on Linux, at launch. A niche case, to say the least.
That said, Intel laid out a number of scenarios where Thread Director will step in to increase general performance and responsiveness of your PC, which all come down to how you personally use your computer. Intel says it will use AI-backed machine learning to constantly train Thread Director to become more effective over time, suggesting that your out-of-the-box performance may get even better as you “break in” your PC, as it were.
Speaking of gaming, for the launch of 12th Generation Core, we’re seeing an upgrade to the integrated graphics processor (IGP) for all “K”-branded chips, now up to Intel UHD 770 Graphics. The UHD 700 series of IGPs is built on Intel’s Iris Xe LP graphics microarchitecture. (As mentioned earlier, Intel continues its long-held tradition of branding non-IGP-bearing chips with an “F” suffix here.) We’ve been told on a separate call with Intel’s engineering team that the new layout will have 32 execution units across all three “K” SKU processors in 12th Generation announced so far (the same as UHD 750 Graphics), while the upgrades include support for up to four 4K displays at once, and adoption of the new HDMI 2.1 cabling standard.
Intel has also engineered a new shared cache architecture in which P-Cores and E-Cores share the same L2 and L3 cache, both of which have been increased in capacity relative to 11th Generation to shoulder the increased load. As for the manufacturing process technology, Intel will base its new 12th Generation stack off what the company is referring to as the “Intel 7 Process,” which, despite its name, contains no 7nm parts. Instead it’s a hybrid of 10nm “Ice Lake” mobile CPU architecture combined with 14nm+++ lithography—much like what we saw in 11th Gen—with the added benefit of the big.LITTLE-style Hybrid Architecture integrated inside.
The combined sum of these efforts? Intel says we should expect “up to a 19% IPC improvement” on 12th Gen vs. 11th Gen, thanks in part to its Performance Hybrid Architecture. We put that part in quotes because you’ll notice that number has a qualifier next to it, labeled “at ISO Frequency.” What’s “ISO Frequency” exactly? So glad you asked!
Going Gaga on Overclocking
Intel took a considerable chunk of its presentation time with us to dig into a category that hasn’t seen much limelight as of late: overclocking.
Intel spoke in depth about all the new bells and whistles that 12th Gen will come with that will supposedly make overclocking both easier and more fine-grained than ever, all depending how deep you want to go. For the “Just gimme a knob to turn” crowd, Intel’s classic Extreme Tuning Utility (XTU) software will be updated to version 7.5, complete with enhancements to the UI and the addition of what Intel is calling its Intel Speed Optimizer (ISO) feature.
This one-click overclocking solution, according to Intel, runs a series of tests that determine the silicon quality and thermal limits of your chip and automatically applies a profile that will push the chip near its peak performance capabilities. On the surface, this doesn’t sound all that different from every other auto-overclocking solution out there. However, Intel says XTU will use specialized data provided by Thread Director, in tandem with its deep-learning AI, to get the best tuning out of your chip.
Whether that’s a lot of marketing hoopla or something that will actually translate to real performance gains, only time and testing will tell. Intel continuously emphasized the amount of headroom that overclockers will find on these chips throughout its presentation, though just how considerable it is will have to be benchmarked to be believed.
Next up, the company announced the refresh of its popular XMP memory overlocking profiler, now in a version dubbed “XMP 3.0.” In this update, Intel promises more control over every aspect of your memory, whether that be DDR4 or DDR5; Z690 motherboards will support either/or, depending on the model.
This includes a new “Dynamic Memory Boost” feature, which Intel says will “deliver intelligent memory overclocking performance on demand,” though, again, the specifics of what exactly all that means for the average overclocker remain to be seen.
When asked about the advantages or risks of using XMP 3.0 and Dynamic Memory Boost during our follow-up call with Intel’s engineers, and the company was quoted as saying:
“There are several advantages to dynamic memory boost, the first being that the timings and speeds from XMP are auto-trained at boot along with the base JEDEC speeds and timings, ensuring a highly stable experience…the entry and exit of Dynamic Memory Boost have been tested repetitively to ensure robustness.
Again, the only way we’ll know what all this means for day-to-day performance and overclocking is by testing it ourselves, and we’ll let you know what we find when these chips go on sale next week.
Alder Lake: Aggressive on Pricing
Earlier this year, we wrote a piece called “After ‘Rocket Lake,’ 5 Things Intel Must Do on Desktop to Get Its CPU Magic Back“, and while the company looks like it may only be three-for-five (as far we can tell, at this juncture), one big thing it got right this time around was costing out 12th Gen competitively against AMD’s Ryzen 5000 Series. (Not an impossible task, truth be told, given how competing AMD silicon has been price-creeping over the past couple of generations; see below.)
While at first that price increase may seem like more of the same for components over the past few years (by which we mean: “It costs how much?”), once you dig into the math, the hidden discounts become more apparent. Though things are a bit all over the place year-to-year, month-over-month in 2021 the US has seen steady inflation that, in theory, would see the price of the Intel Core i7-12700K launching at $423 (using the price of the Core i7-11700K, launched in March of this year, as a baseline).
Instead, we see it here for “just” $409, and that’s with all the supply-chain woes that every manufacturing segment has been feeling since the pandemic. Then we go up the stack to the Intel Core i9-12900K: $589 RCP, to the Core i-9-11900K’s $539 launch RCP.
Sure, a bit over inflation, but also fair considering supply-chain constraints and more cores. The real possible value proposition of the whole thing kicks in once you consider that it’s also got eight added E-Cores and eight more threads, versus what the Core i9-11900K has: “just” eight conventional cores with support for 16 threads. Perhaps sensing an opportunity to let AMD trip over its own pricing hubris, Intel has set up 12th Gen to be an aggressively priced stack of options. That said, you still have to buy a brand-new motherboard, and a premium model at that, to get your new Intel system running.
Z690: A New Platform for Alder Lake
For those keeping count (1…2…3), your numbers are right. In three years, Intel has released three new sockets: LGA 1151 in 2018 for the launch of 9th Generation Core (“Coffee Lake”), LGA 1200 for the launch of 10th Gen (“Comet Lake”) in 2020 (which shared compatibility with 11th Gen), and now LGA 1700 in 2021 for 12th Gen. As competitively priced as these CPUs may be, ultimately they need a motherboard to slot into, and so far we haven’t seen too many options available that would fit nicely into the “budget” segment.
This time around, Intel is launching its next processor line with just one motherboard chipset platform available for purchase: Z690. “Z”-series motherboards are regularly the most feature-rich of the various offerings that follow an Intel chip launch, and also reliably the most expensive. Meanwhile the “B” series has historically covered motherboard options offered in the midrange, while “Hx10” series (for example, H410 or H510) boards bring up the rear for the cost-conscious PC builder, with a few business-oriented “Hx70”-based boards (H470 or H570) also tangling in the midrange to high end.
Despite keeping a keen eye on the rumor mill, we’ve seen nary a peep about “H” or “B” series motherboards coming down the pipeline in the near future other than the MicroCenter leak last week. So that could mean that upgrading to 12th Gen—and LGA 1700 by proxy—will remain at a premium for the indefinite future.
To be sure, the Z690 upgrade does come with perks: support for spanking-new DDR5 memory (up to 4,800MHz), WiFi 6E, and PCIe 5.0 compatibility, just to name a few. However, because we’ve seen/heard only about Z690 thus far, there’s no guarantee that even a desktop with a $264 Core i5-12600K could be built for a reasonable value considering CPU plus motherboard cost…plus, probably, a set of DDR5 DIMMs unless your Z690 motherboard is one of the few that will support DDR4. Let alone what that combined bill might look like if Intel ever launches a 12th Gen Core i3 before 13th Gen desktop CPUs launch in…oh, six months (?), if Intel’s current release cadence keeps up.
The least-expensive Z690 motherboard we’ve heard about thus far comes from Asus (the Prime Z690M-Plus D4) for $189.99, though that floor could lower depending on what more budget-focused brands like Asrock or Biostar put on shelves in the coming weeks. We’ll keep abreast of the platform situation as it develops, but until then, check out our deep-dive on everything we know about Z690 thus far, which will post shortly after this article.
Will Alder Lake Change Intel’s Desktop Fortunes?
Another quarter, and another chip-stack launch from Intel on a version of 14nm (heck, this is the second one of 2021). With an all-new platform on hand, a new core approach, fresh motherboards, new memory, and a new socket, should we be excited about what Intel’s 12th Gen chips and their Hybrid Performance Architecture have in store? Or is it better to keep a cautious eye on the whole thing, given the results we saw during our benchmarks of the so-so Intel Core i9-11900K and Core i5-11600K earlier this year?
With so many new innovations and upgrades on offer, we’re cautiously optimistic. If there was one common thread throughout the messaging around Alder Lake so far, it’s that Intel is done beating around the bush. The company knows it is losing the straight lithography drag race against AMD (read: TSMC) right now, but that hasn’t stopped it from finding new techniques—everything from Dynamic Memory Boost overclocking to Thread Director and more—to squeeze everything it can out of existing process technology as it transitions to 10nm and beyond.
