Intel Coffee Lake Vs. Kaby Lake: A Side-By-Side Comparison

Intel is poised to release its 8th Gen Coffee Lake processors on October 5, which means it’s just about time to start asking if you should get yourself a cuppa the company’s latest CPUs or if you should stick with Kaby Lake. Or if you’re upgrading from a previous generation, whether the extra costs of Coffee Lake (and there are extra costs) make it a better choice than Kaby Lake. Let’s do a little comparison shopping.

Coffee Lake and Kaby Lake are similar in many ways. They rely on the same microarchitecture, with Coffee Lake being effectively (ahem) caffeinated with a slightly refined manufacturing process (14nm++) over Kaby Lake (14nm+), as well as more cores and threads across the board, a different allocation of cache resources, and a few new overclocking knobs and levers.

Each of these chip generations requires a different chipset (100-Series or 200-Series for Kaby Lake, 300-Series for Coffee Lake), and you need to factor that into your cost/benefit equation as well. You will also need to account for the extra cost of a premium “Z” motherboard to overclock the unlocked “K” SKUs. (“H” and “B” motherboards are suitable for the locked models.) Expect to wait until early next year for the value-centric 300-Series H and B motherboards (for Coffee Lake), though. In either case, connectivity options, such as the dual-channel memory support and 16 PCIe lanes from the processor, remain identical for the 300-series motherboards.

MORE: No, Coffee Lake Will Not Run In Z270 Motherboards (And Here's Why)

You don’t visit Coffee Lake for the motherboard, though. Everyone is here for more cores.

The biggest improvement between Kaby Lake and Coffee Lake is the addition of two more cores to Intel’s mainstay Core i3, i5, and i7 families. Intel’s performance projections, which primarily focus on application-level performance as opposed to synthetic benchmarks that scale linearly, appear somewhat modest. But adding 50% more cores to the i7 and i5 series and 100% more cores to the i3 lineup promises to unlock the most performance we’ve ever seen from mainstream Intel processors.

Of course, that’s needed to fend off AMD’s core-laden assault, but the benefits make a splash in the form of 25% more FPS in games (according to Intel) and 45% better “mega-tasking” performance (again, per Intel’s claims). We’ll find out if these CPUs reflect the company’s promises when we load them up on our test bench.

The Coffee Lake lineup has been in development for years, and although the changes to the chips themselves seem designed to parry AMD’s thrust, the relatively early rollout is the only real indicator that Intel is feeling the heat of Ryzen.

For now, though, we want to compare Coffee Lake to Kaby Lake. Let’s walk through the lineup.

Core i7 Coffee Lake vs. Kaby Lake

The i7 series benefits from Hyper-Threading, so the six cores schedule 12 threads in parallel.

Interestingly enough, the Intel Core i7-8700 models feature a drop in base frequency compared to the Kaby Lake counterparts, but this is largely the result of packing more cores into the same package. The additional cores consume more power, which equates to heat. Even though the processor utilizes an improved 14nm++ process, the basic rules dictate that something has to give somewhere.

Intel offsets these lower base frequencies with much-improved boost frequencies, which means the processor will spin up to higher speeds when the additional cores aren’t needed. We’re looking at a 300-400 MHz increase in boost frequencies between generations, which promises snappier performance in lightly threaded applications. Intel’s Turbo Boost algorithms also increase performance at various rates based on the number of active cores, even if all of them are active, but unfortunately the company is no longer providing us with complete Turbo tables in advance of our testing. That’s just a measurement task, so we’ll address that in our review and update here when the NDA expires on Oct 5.

The i7-8700K seems like a more modest upgrade over the i7-8700. It has a 3.7 GHz base frequency, which is a nice bump, but the boost frequency is just 4.7 GHz. The higher-end model also features an unlocked multiplier, but otherwise it’s identical to the i7-8700.

All of this extra power is accompanied by a minor TDP increase, from the i7-7700K’s 91W to the i7-8700K’s 95W, although we imagine extra voltage will magnify that difference when you overclock. The i7-7700 and i7-8700 both sit at 65W, which is a surprising parity given the extra cores.

Coffee Lake i7s also boast support for higher memory speeds, from DDR4-2400 with Kaby Lake to DDR4-2666 with the new processors. Just don’t expect that advantage on the less expensive i3 models, which still top out at DDR4-2400. Intel also added new memory multipliers that let you crank the memory up to 8400 MT/s without modifying the base clock (BCLK). We also get a new on-the-fly memory latency adjustment feature (more on that to come), and speaking of BCLK, the company also added a more advanced PLL Trim feature.

The integrated graphics also receives a slight bump from the Intel HD Graphics 630 clocked at 1,150 MHz (Kaby Lake) to the Intel UHD Graphics 630 at 1,200 MHz (Coffee Lake). Like the previous-generation Kaby Lake processors, you’ll need a license for Windows 10 to use the full graphics functionality, but the difference in graphics capability is minor.

Intel also went with its usual 2MB of L3 cache per core allocation, which nets an increase to 12MB of L3 cache. We’re also assuming that Intel is sticking to its 256KB of L2 cache per core, which also grants us an extra 512KB of L2 cache for the hexa-core models.  

Intel’s i7-8700K sets the tone for what we already expected with the “K” models: It features a price premium compared to the previous-generation i7-7700K. The additional $20 for two extra cores isn’t a big increase, and the move is likely designed more to prevent cannibalization of Intel’s other processors than to offset increased production costs. Meanwhile, the locked -8700 matches its -7700 counterpart’s pricing.

Core i5 Coffee Lake vs. Kaby Lake

Intel made similar changes to the i5 lineup. The company increased the core count from Kaby’s four to six for the Coffee Lake processors, but alas, we still don’t get Hyper-Threaded cores. That means you get only one thread per core, but a 50% increase in cores is welcome.

Again, note the lower base frequencies for the Coffee Lake models, offset by higher boost frequencies. It will also be interesting to compare the mid-range Turbos once we can share that information.

The i5 lineup also features an almost imperceptible improvement to its integrated graphics. Whereas the Intel Core i5-7400 featured Intel HD Graphics 630, which was clocked at up to 1,100 MHz, the i5-8400 and i5-8600K feature Intel UHD Graphics 630 clocked at up to 1,150 MHz. Of course, the change from the “HD” to the “UHD” moniker is more of a marketing exercise than an indication that you receive significantly expanded features, because there aren’t any.

Much like the Coffee Lake i7s, some of the new i5s feature an increased TDP. The i5-8400 and i5-7400 are both at 65W, but the i5-8600K rises to 95W from the i5-7600K’s 91W. The Coffee Lake i5 chips also support up to DDR4-2666 memory speeds versus the Kaby Lake processors’ support for DDR4-2400.

The extra cores also bring a few extra cache slices. The Coffee Lake i5 processors feature a healthy 9MB of L3, compared to 6MB with the previous generation models, as well as 1.5MB of L2 cache.

Again, we see the same pricing for the unlocked models and a somewhat smaller $15 delta between the two locked models.

Core i3 Coffee Lake vs. Kaby Lake

Diving down into the i3 lineup exposes some of Intel’s famed segmentation. We are getting more cores with this generation. The Coffee Lake processors feature four physical cores compared to Kaby Lake’s two, but Intel left i3 memory support at DDR4-2400. That might lead some to opine that the i3 models are just carbon copies of the previous-generation i5 series, albeit with the improved 14nm++ process.

Adding two cores to the i3 models equates to a proportionally larger 100% increase in cores, which you would expect to be reflected in power consumption and TDP specifications. We’ll measure power soon enough, but the i3-8350K’s 91W TDP is a significant jump over the i3-7350K’s 60W rating. We also see a somewhat large jump from 51W with the Kaby Lake i3-7300 to 65W with the Core i3-8100.

Aside from the UHD marketing moniker, the integrated graphics performance should be identical across the two generations, including the clock rates.

Perhaps the most interesting change is the drop in base frequency between the Intel Core i3-7300 and i3-8100. The former is clocked at 3.9 GHz, the latter at 3.6 GHz. (The i3-8350K has the same 4.0GHz base frequency.) That shouldn’t be a deal-breaker, but it’s worth pointing out, particularly because Intel still doesn’t enable Turbo Boost with the i3 lineup. That means you won’t benefit from the higher Turbo ratios we see in the i5 and i7 series. That won’t be as big of a problem with the overclockable i3-8350K--it's expected that users will crank up the clocks--but it will be interesting to see how that impacts the i3-8100 in common desktop applications.

The i3-8100 also sees the smallest L3 cache allocation of the lineup, with a mere 6MB.

Intel is also continuing its new-found practice of offering a “value” i3 SKU that requires an expensive Z-Series motherboard for overclocking. The prevailing opinion is that this approach isn’t popular, but Intel’s continuation of the practice might suggest otherwise.

Stay tuned for our full review when Coffee Lake Launches on October 5.