Core i9-11980HK review: Intel's flagship CPU slugs it out with Ryzen laptops
- 20 May, 2021 20:45
Intel isn’t messing around with its new 10nm “Tiger Lake H” processors for gaming laptops and work notebooks. We’ve already seen how Intel’s 11th-gen Core i7-11800H offers impressive performance over its older 10th-gen sibling and trade blows with AMD’s massively improved Ryzen laptop chips—but how does the new top-end Core i9-11980HK perform?
Today, we’ll give you a taste of where Intel’s Core i9-11980HK lands, but we cannot do so without giving you the lecture. People tend to fixate on a CPU as a monolithic component but they’re not, especially in chips that are built for opportunistic boost clocks based on how much cooling and power they’re given. That’s virtually all of them these days.
Put an Intel or AMD CPU into a bigger laptop with more cooling and power and you tend to get better performance. Drop the exact same CPU into a tiny, thin notebook with less power and cooling and you tend to get lower performance. So always remember to factor in the laptop around your CPU of choice when making any evaluation.
How (and what) we tested
With that out of the way, our preview comes via an Intel-spec’d laptop featuring the top-end 8-core, 11th-gen Core i9-11980HK CPU. The laptop comes with a 16-inch, 2560x1600 resolution display with a 16:10 aspect ratio, powered by an Nvidia GeForce RTX 3080 Laptop GPU with a TGP of 65 watts. It also features a pair of 512GB PCIe 4.0 SSDs and 32GB of DDR4/3200 RAM. Intel’s notebook weighs a relatively light at 4.6 pounds and is powered by a 180-watt power brick.
Although it carries an Intel logo on its aluminum lid and bezel, this system is clearly also manufactured by MSI, and likely based off of MSI’s new Creator 16 series.
Our pile of comparison notebooks comes from the same set of testing results seen in our our performance review of the 11th-gen Core i7 chip. All wield 8 cores—no 6-core or 4-core chips were allowed:
- Asus ROG Zephyrus G14 with Ryzen 9 4800HS, GeForce RTX 2060 Max-Q and 16GB of DDR4/3200. It has a 14-inch screen and weight of 3.6 pounds.
- Asus ROG Flow X13 with Ryzen 9 5980HS, GeForce GTX 1650 Max-Q, 32GB of LPDDR4X/4266. It has a 13-inch screen and weight of 3 pounds.
- Asus Strix Scar G17 with Ryzen 9 5900X, GeForce RTX 3080 Laptop GPU with a TGP of 130 watts and 32GB of DDR4/3200. It has a 17.3-inch screen and weight of 6 pounds.
- Gigabyte Aorus 17G with Core i7-10870H, GeForce RTX 3080 Laptop GPU with a TGP of 105 watts and 32GB of DDR4/2933. It has a 17.3-inch screen and weight of 6.1 pounds.
- Dell XPS 17 9700 with Core i7-10875H, GeForce RTX 2060 Max-Q and 32GB of DDR4/3200. It has a 17-inch screen and a weight of 4.6 pounds.
- Gigabyte Aorus 17G with Core i7-11800H, GeForce RTX 3080 Laptop GPU with a TGP of 105 watts, and 32GB of DDR4/3200. It has a 17.3-inch screen and weight of 6.1 pounds.
All of the laptops are running Windows 10 2H02 19042.928, with one laptop on 19042.985. Every notebook was updated with the latest drivers and BIOSes available directly from the manufacturers before testing, and we benchmarked the systems with each laptop’s highest sane performance fan and performance setting selected. We opted not to run each laptop’s fans at 100 percent as we don’t believe many will do that unless they own stock in a hearing-aid company. (See our Core i7 benchmarks for deeper discussion of power and fan profiles.)
A technical note about boost speeds
Intel actually allows its 11th-generation Core laptop processors to run on anywhere from 35 watts to 65 watts depending on the CPU model, and how the actual laptop manufacturer configures its system.
The Intel laptop’s power level one (PL1) setting was set for 65 watts, while the PL2 was set for a beefier 109 watts. Think of Intel’s power level settings like the gears of a car. This Intel-supplied notebook revs up to its 109 watt TDP for short boosts of 2.4 milliseconds or less when you need maximum performance. Once the load or heat builds up too much, it shifts to the next gear of 65 watts for 56 seconds—or tau—until the heat or power draw once again builds up too much, and then it shifts down again to 45 watts until the work load finishes. The maximum RPM or watts, as well as how long the engine stays there, is basically determined and tuned by the laptop maker.
In our particular suite of tested laptops, the Core i7-11800H was set to a PL1 of 70 with a tau of 64 seconds, and a PL2 of 109 watts and tau of 2.44ms. The Core i7-10870H had a PL1 of 62 watts for 64 seconds, and PL2 of 135 for 2.44ms. Finally the Core i7-10875H’s PL1 is set to 75 watts for 56 seconds with the PL2 set for 135 watts for 2.44ms.
For the AMD, both the Ryzen 9 4800HS and Ryzen 9 5980HS reported PL1’s of 54, with the PL2’s at 65. The Ryzen 9 5900X had both PL1 and PL2 reported at 80 watts. AMD’s CPUs don’t operate quite the same as Intel’s though. Instead the chips intelligently track how hard they can boost based on three key metrics:
- Package Power Tracking, which dictates the maximum power that can be delivered to the CPU socket, measured in watts.
- Thermal Design Current, which is the maximum amount of power the motherboard’s voltage circuits can supply to the CPU measured in amps.
- Electrical Design Current, which is how much power the chip can draw in a short time period. EDC is also measured in amps.
In the same order as above: the Ryzen 9 4900HS was to set to 42 PPT, 51 TDC, 90 EDC. The Ryzen 9 5980HS was set to 42 PPT, 51 TDC, 105 EDC. The Ryzen 9 5900HX was set to 54 PPT, 74 TDC, 127 EDC.
If all this is putting you to sleep, that’s fine. Most people don’t know how many valves their engine packs, or how much boost the Turbo produces. The only thing that matters is what happens when you mash the gas pedal to get on the freeway—so let’s find out.
3D modeling performance
We’ll start with Maxon’s Cinebench R20 benchmark. It’s a popular 3D rendering test that goes faster the more cores you throw at it. Laptops with better cooling and beefier power typically do better, and we see that here where the Core i7-11800H in the thicker Aorus G17 outpaces the Core i9-11980HK in the thinner Intel laptop. It’s not just Intel either, as we see the thinner laptop with a Ryzen 9 5980HS inside also loses to the Ryzen 5900HX in the thicker laptop. Like we said: Laptop design matters!
The real win for Intel here comes when you compare the new 11th-gen Core CPUs to the older 10th-gen parts, which are simply at their limit.
Moving on the popular and free Blender, we render the Barbership Interior benchmark using the default Cycles renderer. The Ryzen 9 5900HX again takes the overall win, which doesn’t surprise us because that Asus Strix Scar 17G and its Ryzen 9 are simply beastly in performance. The test takes about 10 minutes (or more) to run on these laptops. Our Blender test also stresses all cores, so the 11th-gen Core i9 again slightly trails the 11th-gen Core i7 despite the higher model number. What’s particularly impressive is the performance of the Ryzen 9 5980HS in the tiny Asus Flow X13.
We run the V-Ray 5 ray tracing renderer on the CPUs, not the GPU. Like both previous tests, V-Ray also loves CPU cores. The 11th-gen Core i7 actually just squeaks by the Ryzen 9 5900HX chip, with the 11th-gen Core i9 in Intel’s smaller laptop doing pretty well too.
Our last multi-core 3D modelling test is POV-Ray 3.7. This benchmark dates all the way back to the Amiga and can generate beautiful ray-traced graphics based on just a few lines of text. Is it as popular as Maxon’s Cinebench, which is based on its commercial Cinema4D engine, or the open-source Blender that enjoys a large and active user base? Probably not, but it’s still a useful test since POV-Ray does see some use. (It’s in UL’s PCMark 10 test for example.) Ryzen clearly has an advantage here, with the Ryzen 9 5980HS in the thin and light Flow X13 outpacing the 11th-gen Core i9 to the tune of 29 percent. In fact, the older Ryzen 9 4900HS is basically dead even with the 11th-gen Core i9.
We could go back and forth with even more 3D modeling benchmarks but we’d have to generally give the advantage to the burly Ryzen 9 5900HX and Ryzen 9 5980HS. The new 11th-gen Tiger Lake H chips get pretty damn close most of the time, with some results putting Intel dead even or ahead. Coming from the brutal beating that Ryzen has been handing the older 10th-gen Core chips, we’d say that’s a win for Intel.
Not everything is this world uses every core though, so we also look at single-threaded performance using POV-Ray. Where the all-core loads previously favored the big and thick laptops, we now see the thinner laptops actually pull ahead slightly since they are no longer as handcuffed by power and thermal limitations. The 11th-gen Core i9 rightfully pulls past the 11th-gen Core i7 and is nearly even with the Ryzen 9 5900HX.
We also see the older 14nm 10th-gen laptops do relatively well, as neither winds up power or thermal constrained in this benchmark.
Our last rendering test uses Cinebench R20 set to a single thread. The 11th-gen Core i9-11980HK and the Ryzen 9 5980HS lead the way. The Core i9-11980HK also crosses that nice big round number of 600, which is always a nice milestone.
Next page: More Core i9 tests, conclusion
Most everyday consumers don’t really use 3D modeling applications, but they might convert video files. We use the popular and free Handbrake 1.3.3 to convert the 4K open-source Tears of Steel video to the advanced H.265/HEVC codec at 1080p resolution. We report the only thing that we think matters: How long it took to finish the encode.
Having access to more cores increases Handbrake performance and the Ryzen 9 5900HX again flexes much muscle in the thick Strix Scar G17 laptop. For the 11th-gen Core i9 in Intel’s slim laptop, we really think the showdown is against the thin Flex 13 with the Ryzen 9 5980HS inside, and there it’s a tie. Also worth noting: The Dell XPS 17 9700 with the 10th-gen Core i7-10875H weighs the same. Intel has said to expect about a 19 percent efficiency increase in its latest chips, and we see the that with the 11th-gen Core i9 finishing the encode 19 percent faster.
Not everyone uses the CPU for encoding though. Modern laptop CPUs typically have dedicated hardware for converting advanced CODECs, so we tasked the laptops with converting the same file using either Intel’s QuickSync or AMD’s VCE. The 11th-gen Core i7 in a larger laptop finishes the job slightly faster than the 11th-gen Core i9, but the surprise results come from Team Red. We’re not certain why. We would have expected the three Ryzen notebooks to be fairly similar considering all three have the same media engines. But there are key differences in size, weight, and even memory. The Flow X13, for example, uses LPDDR4X instead of DDR4. Nevertheless, this looks like a solid win for Intel.
Many of the previous tests can be boring and difficult to relate to for some people, since all of the chores—though valuable to benchmark—are fairly niche. Adobe is probably a name you’ve heard of though so we looked at the performance of the laptops using the company’s immensely popular Creative Cloud applications. One snag in trying to compare CPUs is Adobe’s reliance on graphics performance. A faster GPU can immensely help some actions in Adobe applications. And since we have different GPUs and at different power settings, it can really make it difficult to judge raw CPU performance.
To remove the discrete GPUs in each laptop from the equation, we simply disabled them and ran the test using each CPU’s integrated graphics. This introduces some variability of its own however as the integrated graphics in these CPUs also differ quite a bit. In fact, the performance of Ryzen’s graphics cores give it a nice advantage.
To give you an idea of how big a difference there is between the visual firepower in our tested CPUs, here’s Geekbench 5.4.1 and its OpenCL test, which measures compute performance of the integrated graphics. Even though the Tiger Lake H chips use Intel’s newest Xe graphics cores, Intel decided to step back from the 96 execution units in the 4-core lower power version that goes into thin and light laptops like the Acer Swift 3X. Tiger Lake H comes with 32 execution units. AMD, however, uses the same die in both thin and light notebooks as well as thicker and heavier laptops.
UL’s Procyon benchmark uses Adobe Premiere Pro. It basically measures how fast each PC exports two 1080p and two 4K files with various filters and color adjustments applied to the videos.
Both of the Ryzen 5000 laptops cross the finish line first, which is a victory for AMD, but both of the 11th-gen laptops are right there as well. It’s an especially good result for Intel when you compare it the older 10th-gen laptops. The thinner 11th-gen Core i9 laptop, for example, offers up an actual whopping 37 percent higher score than the thinner 10th Core i7 in the older XPS 17 9700.
You can argue the results really above really only show two losers for Premiere Pro: The older 10th-gen CPUs indicated by the two light-blue bars. In a world where you would turn off your discrete GPU that’s certainly true, but for a lot of people, the star attraction of a laptop is that Nvidia GeForce GPU inside. Few people who buy powerful notebooks turn off the components. That also plays into Intel’s decision to drastically cut the graphics cores in its Tiger Lake H chips. The company knows most of these chips will be paired with discrete GeForce graphics in laptops so why waste the space?
So to leave you with the full picture, we also ran Procyon’s Video test with the GeForce GPUs enabled, as you can see in the chart below. The result actually pushes the two 11th-gen laptops ahead of the pack despite the Ryzen 9 5900HX being paired with a higher wattage GPU. Both 10th-gen laptops also see a sizeable Premiere Pro encode boost thanks to their GPUs. Where the 10th-gen Core i7-10875H was 37 percent slower than than the 11th-gen Core i9-11980HK with their GPUs disabled, the gap closes significantly to 16 percent with the GPU performance factored in.
We used Procyon’s Photo test for still photography performance. It measures Adobe Photoshop and Lightroom Classic speed in various photo-based tasks. It’s mostly a wash, but the 11th-gen and Ryzen 9 5900HX slightly lead the pack. The 11th-gen laptops also finish roughly 20 percent faster than the older 10th laptops as well.
This test was also performed with the discrete GPU disabled. We’ll mention that both of the 11th-gen laptops actually move out front by a small margin when you turn it back on.
Moving on to applications that get used by millions of people every day, we use UL’s PCMark 10 Application test to measure Microsoft Word, Excel and PowerPoint performance in tasks you do at work. It normally also measures browser performance using Edge but the test failed on all of the laptops. We were still able to measure the individual scores though and the overall win goes to the mighty Ryzen 9 5900HX once again. But once again, 11th-gen Core i9 is right there with it. In a win for Intel, the Core i9-11980HK also outpaces the Ryzen 9 5980HS in the thinner laptop category.
We’d call it a stalemate between the Ryzen 5000 and 11th-gen Core chips. The older Ryzen 4000 and Intel 10th-gen chips aren’t slow, but definitely off the pace to the tune of 15 to 16 percent.
We’ll cap this off with Speedometer, which is designed to measure how responsive web applications are. This time, the 11th-gen Core i9 actually chalks up a real win over the 11th-gen Core i7 by 6 percent and the Ryzen 9 5900HX to the tune of 10 percent. There’s also that giant 27 percent over the 10th-gen Core i7-10875H too.
Our last category is AI Performance. We’ll understand if you just don’t care because “AI” is the buzzword you hear when a CEO gets excited about creating more intelligent chat-bots so the company can lay off even more people. But consumers are finally seeing application of advanced AI techniques to make far more nuanced and intelligent decisions than before. Intel dominates in this area for now thanks to a strong focus on boosting performance in tasks that deploy deep learning.
Our first benchmark is Nero’s Score 1.08, which a new free test that attempts to measure CPU performance at recognizing and tagging your photos for you. Your smartphone might offer AI image sorting but that doesn’t help you if you’ve got 50,000 photos on your PC. Nero Score is based on its commercial AI Photo Tagger that we’ve used to measure AI performance before, but this benchmark uses a smaller photo set and at lower resolution than tests we’ve done. That makes it better as a CPU test.
The 11th-gen Core i9 is a stunning 45 percent faster than Ryzen 9 5900HX. It’s also 73 to 90 percent faster than the older 10th-gen chips as well. Whoa.
Our next test uses Topaz Lab’s Gigapixel AI 5.2.2 to increase the resolution of a 10MP DSLR photo taken a decade ago. The results are generally very impressive and beyond what’s capable with simpler algorithms you’re probably used to. It’s also quite compute intensive. The Core i9-11980HK finishes the job in 47 percent less time than the Ryzen 9 5900HX and 68 percent faster than the 10th-gen Core i7-10875H.
To show you how fast this field is moving: Between our Intel 11th-gen Core i7 testing late last week and now, Topaz Labs has already released a newer version with newer models. Our testing reflects that.
The massive speed gains are due to Intel’s push to support AVX512 in its CPUs. With Tiger Lake H, the company now offers CPUs with AVX512 or its Deep Learning Boost technology in everything from ultra-light laptops to heavy-duty servers. Applications that use it get massive performance benefits.
AMD has argued that yes, that’s neat, but very few applications support AVX512 right today so who cares? That’s a fair point as probably fewer than a dozen consumer applications play nice with it for now. At the same time, these practical applications that yield very tangible benefits indicates that this field will only continue to grow, so consider that before you buy your next laptop.
What, no gaming?
We aren’t covering gaming performance today. While the Intel reference laptop has a 65 watt GeForce RTX 3060 Laptop GPU, we’re not entirely comfortable making comparisons to the other laptops and any attempt at this point feels premature.
Ideally, we’d put similar-sized laptops with with the same model GPUs and very close power ratings against each other to determine the better CPU for gaming. That isn’t here today but if you want our forecast: Tiger Lake H-based laptops will either be faster or so competitive with Ryzen, it just won’t matter.
The truth is most gaming is still largely driven by the GPU performance. For example, you’ve seen the 11th-gen Core i9 offer 20 percent to 50 percent more performance than the older 10th-gen in CPU tasks, but in gaming with the same GPUs they’ll probably be a lot closer than you expect.
If you’re looking for buying advice today for gaming: Pick the biggest, fattest GPU you can afford and don’t sweat too much whether it’s a 10th-gen, 11th-gen, or Ryzen 5000 chip. We’d take a 10th-gen Core laptop with an RTX 3080 over an 11th-gen (or Ryzen) laptop with an RTX 3060.
Let’s close things out by running the older Cinebench R15 benchmark on everything from a single thread all the way up to 16-threads. We perform this task in CPU evaluations because it helps us visualize the overall performance of a CPU on both very lightly-threaded tasks and very heavily-threaded tasks, as well as anything in between. You can see the results of all the laptops tested in the graph below.
Intel’s problem with its 10th-gen CPUs began when the Ryzen 4000 appeared and offered as good (or even better) multi-threaded performance on much thinner laptops. Ryzen 5000 twisted the screws by offering even more multi-threaded performance while greatly increasing its lightly threaded performance.
The first two Tiger Lake H laptops we’ve tested change that narrative. What we’ve seen so far doesn’t make the Tiger Lake H the clear-cut winner against Ryzen 5000 notebooks, but performance between the rivals is close enough now that it’s not really that big of a deal.
For example, here’s how the Intel laptop performs in Cinebench R15 compared to the Asus Flow X13, showing performance as a percentage difference from one to 16 threads. And yes, we think here it’s fair to make the comparison even though the Intel laptop is 4.6 pounds while the Asus ROG Flow X13 is three pounds. That’s because both laptops are almost the same thickness measured from the keyboard to the base. (The Flow X13 also features a 35-watt GeForce GTX 1650 Max-Q vs. the 65 watt GeForce RTX 3060 Laptop in the Intel laptop.)
With that said, we’re pretty certain the Core i9-11980HK uses more power than the Ryzen 9 5980HS (which we’ll visit at a later date) but both CPUs are practically in lock step from very light loads to very heavy loads.
The big win for Intel is clearly over its aging 10th-gen CPUs, as shown in the graph above. Although AMD’s Ryzen 9 battles the 11th-gen Core i9 to a standstill, when we perform the same comparison against a similarly sized laptop with the 10th-gen Core i7-10875H in it, it’s an absolute slaughter. You may scoff because the 10th-gen isn’t a Core i9 model, but we’ve done enough testing of the 10th-gen Core i9 to tell us it wouldn’t have changed the results by much at all. That’s because the 10th-gen CPUs were at the end of the road for Intel’s long-in-the-tooth 14nm manufacturing process.
Clearly, Intel’s 11th-gen Tiger Lake H is a force to be reckoned with. Does it defeat its dreaded foe, AMD’s Ryzen 5000? No. But it deserves to be in the conversation at least. We couldn’t say that about the 10th-gen chips. And once that conversation gets started, Intel can throw out its advantage in having PCIe gen 4 lane GPUs, PCIe gen 4 SSDs and Thunderbolt.
So no, Intel and AMD’s raging battle for laptop supremacy isn’t over yet—but at least it’s interesting again.