Intel’s 10th-gen Comet Lake H is easily the best Core i7 available today in a gaming laptop. In fact, we’d be bold enough to say that 10th-gen Core i7 Comet Lake H is easily the best Core i7 we’ve ever seen in a gaming laptop.
With the 8-core, 16-thread Core i7-10875H, Intel’s finally giving gamers and power users the cores they want in a Core i7. Previously, Intel sequestered 8-core chips behind that First-Class curtain accessible only to Core i9 buyers. With Core i7-10875H, such luxury is trickling down to Economy Plus users.
Intel isn’t doing this out of the goodness of its heart, though. A little 7nm chip called Ryzen 4000 is looming over every single result here, having sent shock waves through the laptop world when it made its debut last month.
Our Comet Lake H news story by Mark Hachman provides more detail, and we’ve lifted out the most important part in this chart listing the specs of the new lineup. The Core i5 doesn’t change much other than clock speed increases. The same goes for the two mid-range Core i7 chips, which get modest boost clocks.
For launch day, we didn’t have access to the top-of the line Core i9-10980HK. Maybe that’s for the better, as the Core i7-10875H is likely the sweet spot of the lineup.
How we tested
Testing a mobile CPU is not at all like testing a desktop CPU. In a desktop, the reviewer can control what GPU, what SSD, what RAM, and what cooling is used, and test apples to apples or darn close to it.
Every laptop, on the other hand, is a custom-built platform. The closest anyone can get to apples-to-apples testing is on that rare occasion when the vendor offers two different CPUs in the same rig. For instance, we compared Intel vs. AMD CPUs in Acer’s Predator Helios 500, the sibling laptops being nearly identical otherwise.
When you test a laptop CPU, what little you can control are comparable size and weight. Larger and heavier laptops can typically offer more space for cooling and power. It’s not fair to compare it to a thin-and-light laptop, which will face more challenges with ventilation and likely throttle performance to keep things cooler.
Even so, there is real value in looking at CPU performance if the benchmarks are mostly limited to the CPU.
For our test, we used Gigabyte’s updated Aero 17 HDR. With a gorgeous 17.3-inch, UHD 4K screen that can hit HDR400 Vesa specs, the laptop also comes with a Core i7-10875H, 16GB of DDR4/2933 in dual-channel mode, and Nvidia’s new GeForce RTX 2070 Super Max-Q GPU.
For our tests we set the laptop to its “Gaming” mode for fan speed controls and also set the CPU and GPU to their “max” setting. We stuck mostly to CPU tests to measure the 10th-gen Core i7 Comet Lake H CPU performance. We also ran some GPU tests, as we had high hopes for Nvidia’s GeForce RTX 2070 Super Max-Q GPU.
10th-gen Core i7 Comet Lake H Performance
We’ll kick off our tests with Maxon’s Cinebench R15. It’s a multi-core benchmark built off of Maxon’s older engine used in its professional Cinema 4D test. While 3D modelling isn’t exactly an everday task (even though Cinema 4D is integrated into Adobe’s After Effects and Premiere products), but we like Cinebench R15 for being consistent and reliable at illustrating multi-core performance loads.
The result for the newest 10th-gen Core i7 is quite good. As you can see, the Core i7-10875H hangs right with the 8-core crowd and comes fairly close to the premium Core i9-9980H in the MSI GE65 Raider. And yes, there is that Ryzen 9 4900HS too. While AMD fans will want to interrupt the discussion to chest-thump, save it for the end.
One issue with Cinebench R15 is its vintage. It’s now seven years old, which is getting up there in benchmark years. It’s been useful on laptops only because laptops haven’t taken the leaps desktops have in core count. Of more value these days is its replacement, Cinebench R20, which came out last April. Besides updating the test with Maxon’s newer Cinema4D engine, R20’s workload takes three times as long to run and adds AVX512 to the equation, straining modern CPUs harder. Also, by taking longer to run, you get a better idea of how long a CPU’s boost clock can hold up under heavy loads.
As you can see above, the new 10th-gen Core i7 jumps slightly in front of the Core i9-9980H chip in the harder-to-run Cinebench R20. The fact that a Core i7 is faster than a previous-gen Core i9, even by a little bit, is a good sign for consumers.
Yes, AMD fans, we see that Ryzen 9 4900HS up there near the top too, so we’ll say it before you burst. Of course, Intel can say, wait for our Core i9.
Intel may not like the multi-core performance in Cinebench, but it certainly likes single-core tests. We’ll skip Cinebench R15 (which you can see in our review of the Gigabyte Aero 17) for Cinebench R20. The test is the same, but rather than using every core on the CPU, the test only uses one. For all our love of multi-core, a huge swath of software still depends more on single-core performance.
It’s a win here for the 10th-gen Core i7-10875H, as it leads an impressive class of “9” CPUs—everything from the Core i9-9900K in a laptop, to the Core i9-9980HK and yes, AMD’s Ryzen 9 4900HS. The margin is slim; we’d bet most people could not tell the difference in single-core performance among any of these CPUs. Still, that high-boost clock of the new 10th-gen chips is very real indeed.
This pattern, by the way, is mostly replicated through a Costco-sized suite of CPU-focused tests that we ran. We’ll show just a selection, specifically V-Ray and the newer V-Ray Next. Both are professional rendering engines that see use in actual movies, films, and commercials.
The Core i7-10785H can’t beat that Ryzen 9, but remember it’s a Core i7, not a a Core i9. Its performance against the CPU it’s intended to replace is quite good. In V-Ray Next, the 10th-gen Core i7 easily dispatches the 9th-gen Core i7 and the 8th-gen Core i7 with their “mere” six cores.
If you want even more rendering tests here’s POV-Ray 3.7. POV-Ray 3.7 is an Amiga-vintage test updated for modern times. Again, 10th-gen Core i7 does reasonably well compared to its 9th- and 8th-gen siblings and even the Core i9-9980H. There’s Ryzen 9 4900HS on top, though.
ln the 10th-gen Core i7’s defense, when you look at the single-threaded performance, it’s clearly pretty handy to hit 5.1GHz on boost on lightly threaded work.
One issue with most canned benchmarks is the short time they often take to run. Sometimes you really want to heat up a laptop to see if it wilts under pressure. Our test using the older HandBrake file conversion utility did that when H-class laptops had four cores. As core counts have risen, the task that used to take 45 to 60 minutes now finishes in just over 20 minutes. That’s still long enough for most laptops to crank down clock speeds to keep from overheating.
When pushed with HandBrake, the 10th-gen Core i7-10875H rallies. We found the Gigabyte Aero 17 to be just as fast as MSI’s GE65 Raider with its Core i9-9980H. It also, of course, leaves those older 6-core CPUs behind. Ryzen 9 4900HS is still faster—we hear you, AMD fans.
What about gaming performance?
We know, you want to see the gaming performance of the 10th-gen Core i7 chip. We’ll bring up what we said earlier: You can’t separate GPU from CPU in a laptop. And because gaming is heavily influenced by the GPU inside (as well as the cooling of the laptop), it’s pretty hard to draw detailed conclusions.
For instance, we ran Quake II RTX—a fully path-traced version of Quake II—on four modern laptops with varying GPUs and CPUs. Although the test gets a little bump from higher clock speeds, the GPU with more ray tracing hardware inside wins.
Here’s the result from Rise of the Tomb Raider, which is run in DirectX 11 mode for legacy support reasons. The test factors in GPU and CPU to an extent, but again, the big, hairy GPUs easily win this one.
It’s a pretty good sign for Nvidia’s GeForce RTX 2070 Super Max-Q, which can beat some RTX 2080 Max-Q implementations. However, a full-fledged RTX 2080 in a big, fat laptop is always going to win.
We ran some synthetics, such as 3DMark Time Spy’s CPU test, which uses an open-source physics engine to calculate theoretical CPU performance.
The test technically puts the Ryzen 9 4900HS and the beefy big gaming laptops ahead of the Gigabyte Aero 17 and its 10th-gen Core i7 chip. However, we suspect the 10th-gen Core i7 will do exceptionally well in games, thanks to its high clock speed in light loads.
Integrated graphics encoding performance
Modern laptop CPUs pack integrated graphics chips (IGPs) which mostly go unused for gaming or content creation—that’s the job of the discrete GPU, if there is one. The integrated GPU may sometimes step in for video encoding, though.
While the graphics engine in the new 10th-gen Core i7 isn’t actually new, we did want to see how Intel’s QuickSync performance matches up against AMD’s new Video Coding Engine.
We decided to use the neutral ground of HandBrake 1.3.1. The popular and open source encoder supports CPU encoding, as well as Intel’s QuickSync, AMD’s VCE and Nvidia’s NVENC. We took the open-source Tears of Steel 4K movie and encoded it using the H.265 preset for 1080p and 30 fps.
First up is a CPU encode. The winner is AMD's Ryzen 9 4900H, but the 10th-gen Core i7 is fairly close behind. The 6-core CPUs slide in much later, with the 9th-gen Core i7-9750H ahead of the older 8th-gen Core i7-8750H.
The previous test is pure CPU. Let’s see what happens when we pit UHD QuickSync vs. AMD’s Zen 2 VCE performance. Note: the Omen X XS drops out of this test, as HP turns off the integrated graphics and runs the discrete graphics full-time.
It looks like AMD’s VCE combined with the Zen 2 cores again squeezes out a win—though a very close one. We actually thought the two Intel laptops would be far closer because the dedicated hardware in the IGP should be doing the work, but it looks like core count still matters, too.
We also ran HandBrake with Nvidia’s NVENC engine. That’s the dedicated encoding hardware in the Turing-based RTX cards. Again, we selected H.265 4K to 1080p/30, but we selected the NVENC codec. We expected all the NVENC performance to be the same because all of the RTX cards use the same NVENC engine. But just like the IGP performance, core count and clock speeds still matter.
Our last test that likely slots under encoding or transcoding uses Topaz Lab’s Video Enhance AI app to convert a 90-second, 720p video (shot on an old Flip-style video camera). The app uses machine learning and Intel’s OpenVINO to perform a frame-by-frame analysis of the file and upscale it to 1080p resolution. The software doesn’t currently use Intel’s DL Boost, and when set to “CPU” seems to lean very heavily on the CPU cores. The CPU-focused test basically loads up the CPU cores to 100 percent for more than five hours.
We see the Ryzen 9 4900HS predictably ahead of 10th-gen Core i7 by a small margin. What confuses us is seeing the Omen X 2XS coming in just behind the 8-core 10th-gen. We would have expected it to be much farther behind, like the Core i7-8750H, but oddly it’s not.
We then ran Topaz Video Enhance AI on a GPU, upping the conversion resolution to 4K. This is a realistic task for someone who wants to take a 2009-era Flip video and convert it to 4K for display on a TV.
Note that while it takes in excess of five hours to upsample it only 150 percent using only the CPU, the Gigabyte Aero 17, with its 10th-gen Core i7 and GeForce RTX 2070 Super Max-Q, finishes the job in 85 minutes with an upsample of 400 percent. If you’re interested in diving deeper into this Topaz Video Enhance AI rabbit hole, read Joel Hruska’s obsession with up-sampling Star Trek: Deep Space Nine over at ExtremeTech.com.
Let’s close this off where we started: The 10th gen Core i7 is easily the best mainstream Core i7 ever seen. We always like to prove it with a few more charts.
Here’s a task where we take Cinebench R15 and run it using from 1 to X threads available on a CPU. We like to do this because it’s an easy way to visualize where the strength of a CPU may lie. If a chip gets high clocks on light loads, but falls apart on multi-core loads—you can see it.
First is the percent difference, between the new 10th-gen Core i7-10875H and the previous Core i7-9750H, the darling of many gaming and power-user laptops. As you can see, the new 10th-gen Comet Lake H simply crushes its older cousin across the board, from light loads to heavy loads.
We also look at how the new 10th-gen Core i7 Comet Lake H does against the older (and pricier) 9th-gen Core i9-9880H chip. As you can see, the new 10th-gen Core i7 easily wins in light loads against the Core i9, but the Core i9 earns its pay on heavier multi-threaded loads.
Of course, the 7nm elephant in the room continues to be Ryzen 4000—namely the Ryzen 9 4900HS. Although it’s not perfectly fair to pit a Ryzen 9 against a Core i7, we should note the Ryzen is the low-power HS version, not a full-tilt H version. It’s also locked into a 3.5-pound laptop, which means limited cooling compared to the 5.8-pound Gigabyte Aero 17. What’s more likely to matter is how Ryzen 7 shows against Core i7.
For now, we don’t want to lose sight of the fact that this is a red-letter day for Core i7 laptop CPUs everywhere. To help you visualize this, we’ve mapped out Intel’s heavy-duty CPUs from the 1st-gen Arrandale dual-core chip to today’s 10th-gen Comet Lake 8-core chip. We can say, without a doubt, that the Core i7 has come a long way. Even when its was stuck at quad-core for too many generations, it’s been moving briskly of late.