INTEL’S HOT NEW CHIPS EXPLAINED
Under the hood of Intel’s 8th Generation processors
INTEL’S 8TH GENERATION core processors have landed. They’re the most exciting mainstream CPUs from Chipzilla in at least five years. But also the most confusing. In fact, Intel’s 8th Generation Core isn’t really a generation of CPUs at all. It’s more a collection of products with common branding than a distinct generation of processors with something approaching a homogeneous feature set.
Eventually, 8th Gen Core will contain CPUs from three generations of processor design, and be manufactured on two silicon production nodes. But even with three generations and two production nodes, 8th Gen Core chips don’t include Intel’s new multi-core chips that range from 10 to 18 cores. Those fall under the 7th Gen banner. Intel never makes it easy to understand its product range.
If Intel’s marketing can be utterly baffling at the best of times, 8th Gen Core remains really good news for the most part. For starters, we’re looking at the first major step forward in mainstream desktop CPU performance from Intel for a very long time. In simple terms, you’re getting more cores at every mainstream price point. Hurrah! Mobile performance and battery life are on the up, too.
Where the confusion comes in is that, unlike previous generations, where being a part of a specific Generation of Core processor meant something in terms of features and capabilities, the simple fact of a CPU being 8th Gen doesn’t tell you much. Instead, you’re going to have to stay on your toes to keep up with the pros and cons of Intel’s new range.
With that in mind, we’ve drawn together everything you need to know about these new CPUs. From poverty priced entry-level chips to mobile masterpieces and desktop killers that set new performance standards at certain price points, we’ve got the full low-down. So, now you’ll know what to expect, where the best bang for buck is to be found, and which chips fit best with your PC usage. Let it rip!
CHRONOLOGICALLY SPEAKING, Intel’s 8th Generation CPUs began where it so often does these days: with a mobile CPU announcement. But right off the bat, something wasn’t quite right. A quartet of new “U” Series mobile processors debuted the 8th Generation nomenclature, including 8000 Series branding. But they were based on neither a new silicon production node nor a new processor architecture. Seriously?
That’s worth repeating. Intel is calling the chips 8th Generation, but they are manufactured on 14nm production tech, which dates back to the Broadwell generation in 2014. Since then, Intel has revised its 14nm offerings with the Sky Lake and Kaby Lake architectures. So, this would be the fourth new CPU family on the 14nm node, although Intel characterizes it as a revved-up “14nm+” process, with improved transistor fins, increased channel strain, and other tweaks that add up to a claimed 12 percent performance increase from process improvements alone.
That’s unprecedented for Intel. Not long ago, Intel used to design CPUs to conform to its so-called Tick-Tock cadence of chip creation, in which the aim was to crank out a new architecture, then shrink it to a smaller production node in successive years. But in early 2016, Intel announced it was ditching Tick-Tock in favor of Process, Architecture, Optimization. In other words, in year one, Intel would shrink an existing CPU to a smaller silicon production node. In year two, it would release a new architecture on that new node. And in year three, it would revise and optimize that new architecture.
The move was an implied admission that Moore’s Law is toast. Along with the rest of the integrated circuit industry, Intel is increasingly bumping up against the laws of physics. Shrinking its transistors every other year is no longer a goer. But that much we already knew before 8th Gen Core arrived. The surprising thing about those first mobile processors is that they don’t even conform to Intel’s new Process, Architecture, Optimization cadence. In fact, they’re a mildly tweaked revision of the existing 14nm Kaby Lake architecture.
Arguably, of course, that’s just details. What really matters is that with the new 8th Gen U Series mobile CPUs, Intel replaced a quartet of dual-core models with quad-core CPUs, and is doing so at slightly lower price points. The one downside involves clock speeds. Peak Turbo frequencies have been maintained or even improved, but base clocks have reduced a little. That reflects the fact that doubling the core count and maintaining both frequencies and power consumption, even with a tweaked 14nm+ process, is a non-starter.
As for specifics, at the bottom end of the relevant U Series range, Intel used to offer the Core i5-7260U for $304. For that you got two cores and four threads clocked at 2.2GHz base and 3.4GHz Turbo. Now, Intel will sell you the Core i5-8250U with four cores and eight threads, a base clock of 1.6GHz, and a top Turbo speed of 3.4GHz. And it will set you back just $297. Nice. At the other end of this U Series scale used to be the Core i7-7660U, a dual-core, quad-thread 2.5/4.0GHz model for $415. Its replacement is the Core i7-8650U, a quadcore, eight-thread puppy rocking 1.9/4.2GHz frequencies, and yours for $409.
If that’s the good news on the mobile side, the bad is that it only adds to the incredible complexity and confusion of Intel’s current product offerings. Right now, Intel lists no fewer than 144 individual mobile CPU models for sale from various generations of CPU. Just mobile CPUs, you’ll note, and all based on the full Core architecture. Of those 144 models, all are either dual-core or quad-core. The upshot is that much of Intel’s branding has become meaningless.
There are Core i5 mobile CPUs with two cores, for instance. There are Core i5 mobile CPUs with four cores. But some of them have Hyper-Threading and support for eight threads. Others do not. Meanwhile, customers are expected to decipher an impenetrable product suffix, with anything up to four numerals and two letters. Frankly, if you set Intel the task of creating
a product lineup expressly designed to beat customers into submission with the sheer, unfathomable complexity of it all, it could scarcely be any different from the array of similar mobile CPU models it does offer.
Despite all that, the new chips are still very welcome. They shift a whole class of notebooks instantly from dual to quad-core processing. Another 8th Gen novelty first seen with these U Series mobile chips is the debut of new branding for Intel’s integrated graphics. Gone is the old “HD” moniker; in comes “UHD.” The branding change reflects an increased emphasis on 4K or UHD resolutions. To achieve that, the new chips now natively support UHD resolutions via HDCP 2.2 content encryption and DisplayPort 1.2a connectivity. However, beyond those 2D video upgrades, the 3D graphics architecture is unchanged. That’s significant. But hold that thought while we segue deftly to the desktop and the CPUs that arguably best define this new generation of chips. We give you Coffee Lake, the family of CPUs that we’d argue is the true 8th Generation Core architecture.
Once again, it’s based on 14nm technology. And once again, Intel is claiming further revisions and advancements for its 14nm transistors. This time around, Intel is calling it 14nm++ and making some fairly bold claims. Overall, Intel says 14nm++ allows for around 24 percent higher drive current for 52 percent less power than its original 14nm process. What hasn’t seemingly changed is much by way of architectural improvements. Intel isn’t making big claims for things such as percore performance. Nor has it added to the 16 native PCI Express lanes connecting the CPU to peripherals, such as graphics.
Instead, the really big news with Coffee Lake desktop chips is that they introduce the first increase in core count for mainstream Intel desktop processors since the Nehalem architecture of nearly a decade ago. Where once you got two cores, you now get four. And where you got four cores, you’ll now be getting six. And you’ll nearly, but not quite, get all that for the same money as before.
Intel has initially launched six CPUs, two each for Core i3, i5, and i7. The Core i3 models now offer four cores, while the Core i5 and Core i7 now pack six. If you’re wondering how the latter are differentiated, it’s via Hyper-Threading. Only the Core i7 models get that, so only the Core i7s are 12-thread chips. For the record, the Core i3s don’t get Hyper-Threading. Nor, incidentally, do they have the Turbo frequency boost mode. You get the base clock, and that’s it.
The exception to that is the unlocked “K” model in the Core i3 range, which is fully overclockable via the CPU multiplier. In fact, there’s an unlocked model of all three versions. What’s more, Intel has improved the overclockability of the K series models with more granular per-core overclocking. What it hasn’t done, unfortunately, is ditch the cheapo thermal interface material (or thermal paste) that connects the CPU die to the processor lid. Soldering the lid to the die would be better for performance and overclocking, but also that little bit more expensive to manufacture.
Regards pricing, the two Core i3 chips and the entry-level Core i5 and Core i7
are all the same price as their 7th Gen progenitors, while the top Core i5 and i7 processors are a little more expensive. The other major change involves clock speeds. Peak Turbo speeds have gone up for the Core i5 and Core i7 chips, while base clocks are down across the board. But the base clock doesn’t tell the whole story. For example, Intel rates the new six-core Core i5-8400 at 2.8GHz base clock. That sounds disappointing, even if it is only 200MHz below the 3GHz base clock of the earlier quad-core Core i5-7400. In practice, however, we found the 8400 runs at 3.8GHz under load across all six cores. That’s just 200MHz off its top 4GHz Turbo speed.
That aligns with Intel’s rating for the 8400 with all six cores loaded. You can refer to our table for the details, but thanks to the added cores, it’s more important than ever to consider how fast each chip will run with all its cores under heavy load. That said, even the slowest of the new CPUs hits at least 3.6GHz with all cores loaded, while the top Core i7-8700K is rated at 4.3GHz with all six cores number crunching. In other words, Intel really is giving us the best of both worlds with these new processors. We’re getting more cores, and clock speeds have only fallen off by a very small amount.
Less impressive, depending on how you look at it, are the integrated graphics capabilities we alluded to earlier. Beyond the change in branding, there’s little to report. Intel hasn’t revised its 3D graphics architecture for Coffee Lake. The “new” UHD 630 graphics found in the new CPUs is the same in 3D rendering terms as the HD 630 in the 7th Generation Kaby Lake chips.
Is this an indication that Intel is finally giving up on integrated graphics for what you might call serious gaming? That’s hard to say. What is for sure is that Intel’s integrated graphics have never been any good for serious gaming, and even fall short for casual gaming with older games. There’s a strong case for arguing these relatively high performance chips would be better off without the on-die graphics taking up any space at all. The Core i5 and i7 models, in particular, will almost certainly be paired with discrete graphics cards.
With that in mind, Intel could either make the six-core die smaller and cheaper by removing the graphics, then pass the cost savings on to consumers, or add more cores and create an arguably more logical and appealing CPU range, with four cores at the bottom on the i3 models, six cores for i5 chips, and eight cores for i7s. It wouldn’t surprise us to see a hierarchy something like that in future generations.
While we’re talking graphics, there’s the question of gaming performance. For
“For now, the number of games that make good use of more than four CPU cores registers between slim and none.”
Now with six cores, but also that pesky integrated graphics. as long as we can remember, true multithreaded gaming has been just around the corner. It’s a bit like cold fusion power, AI, or a cure for cancer. It’s always just over the horizon, but never seems to get closer.
Admittedly, with the arrival of new graphics APIs in the form of DirectX 12 and Vulkan, you could argue that the longpromised multithreaded future for gaming is now tangibly closer. And that may be true, but for now, the number of games that really make good use of more than four CPU cores registers somewhere between slim and none. With that in mind, what to make of the extra cores in Coffee Lake CPUs? For our money, it’s good news however you slice it. If you’re a many-cores denier, and you think four cores are plenty for gaming, the new chips get you a proper quad-core Intel desktop processor for just $117. If, like us, you think that six cores are a solid futureproofing bet, that’s now a mainstream option from Intel.
Of course, while we’re patting Intel on the back for finally delivering some more mainstream processor cores on the desktop, it’s only fair to give its main— and, frankly, only—rival AMD the last word in all this. It’s easy to imagine Intel’s new-found largesse failing to materialize had it not been for the sudden uptick in competition provided by AMD’s exciting new Ryzen processors.
With up to eight proper CPU cores in its own mainstream CPU socket, Ryzen packs a serious multithreaded punch. It’s also much more competitive in terms of single-threaded throughput than AMD chips of late. So, while it’s true to say that Intel almost certainly would have increased the core count of its CPUs of its own accord eventually, it’s unlikely to be a coincidence that it’s happening just when AMD upped its game. Indeed, the argument that Intel’s moves of late have been a direct reaction to increased competition from AMD only looks more convincing when you consider the spectacular decision to increase the core count of its enthusiast-class desktop processors from 8 to 18 in a single generation. There’s no way Intel would have done that without the AMD resurgence.