Intel Core i9-12900HK Review
- 2022-03-15 20:51
- Recompile By Brian - CPU Review
Core i9-12900HK is Intel's newest and fastest CPU for laptops. In this review we'll be comparing it to a bunch of other mobile products, including the monster Apple M1 Pro, so we'll investigate Intel's bold claims to be Fastest mobile CPU.
We'll also compare the 12900HK with the more mainstream Core i7-12700H we reviewed, and the new AMD Ryzen 9 6900HS, among others.
Besides the i7-12700H, the Core i9-12900HK increases the clock speed simply: same core configuration, with 6 P cores and 8 E cores for a total of 20 threads, but the clock is now 100 to 300 MHz higher. .
For example, the 12900HK can peak at 5.0 GHz on the P cores, compared to the Core i7's 4.7 GHz. We've seen a slightly improved Core i9 over the Core i7 in the past, so performance stays slightly better within the same 45W wattage limit, but we'll have to judge that in the future. test today.
The test system for today is the MSI GE76 Raider. We've tested previous iterations of this MSI laptop, and it's clearly designed for high-end gaming on the go. Now it includes not only the 12900HK, but also Nvidia's flagship laptop GPU, the GeForce RTX 3080 Ti with a 150-175W power range.
Whether this type of system is right for you depends on your needs, but I wouldn't say it's designed for portability. As a 17-inch chassis that weighs more than 6 pounds without taking into account its massive capacity block, the GE76 is a desktop replacement that's always connected to wall power and only occasionally moved.
It has a 99.9 Wh battery, but the battery life is pretty bad and I don't want to use it on battery. I prefer thinner and lighter systems when it comes to laptops, but obviously these are also popular and geared toward different needs than everyday portability.
As for the other hardware, the GE76 is equipped with 32GB of DDR5-4800 memory in a dual stick configuration, which is what we've seen on all 2022 gaming laptops so far. In this configuration, there's a 17.3-inch 1080p 360Hz display although it's also available in 1440p and 4K options.
Our testing today was conducted in two power configurations: the long-term PL1 power limit of 45W, Intel's default specification for these chips as well as 75W, which is an increased power rating. Common booster seen on many laptop vendors.
They will be compared apple to apple with other laptop chips running in the same configuration, which gives us the best look yet at how these CPUs compare in a fair benchmark battle.
If we were to compare one chip at 75W with another at 60W, we really only want to compare which particular laptop design can send the most power to the CPU, not which CPU actually performs better, which is why we test like this. Boost PL2 power is set to Intel’s default of 115W in both instances.
Tests like this also give you the most useful information that can be applied to a variety of laptops. If you're buying a thinner and lighter system, the 45W figure will matter most, while the numbers after a more powerful system with more cooling capacity will be 75W.
About the results!
In the multi-threaded Cinebench R23, when the Core i9-12900HK was capped at a default of 45W, it was the fastest CPU in the ranking, with a score of 13400 points. This allows it to outperform the Apple M1 Pro, although Apple's chip is consuming around 35W of pack power, not 45W, which is a trend you'll see on most charts. However, the 12900HK is the winner here with 8%, and so it's also 15% faster than AMD's new 35W CPU.
What probably surprised me the most was the amplitude between 12900HK and 12700H. There's really not much separation between these parts, only a 3% performance increase in favor of the Core i9.
Both parts have the same number of cores and are operating within the same power limit, so the only difference is binning and optimization, which doesn't seem to improve performance significantly. With that said, the performance increase over the 11980HK from the previous generation is huge, a very solid 23% increase.
At 75W we are seeing a continuation of what was shown at 45W, however the gap between the 12900HK and 12700H has now dropped to nothing, at least with the models we tested. This gives the Core i9 a 16% lead over the Ryzen 9 6900HS, as Alder Lake can deliver higher performance-per-watt in the higher power range than Rembrandt, but it doesn't exactly scream at buyers that the Core i9 is a worthwhile choice compared to Core i7.
A larger performance difference between the 12900HK and 12700H was observed in the single-threaded test. With the 12900HK hitting 5GHz, it's 5% faster than the 12700H from a 6% increase in clock speed difference. That's pretty much as expected. Alder Lake is pretty much faster than the M1 Pro here, taking a huge 25% performance lead in single-threaded workloads like this, and 19% faster than AMD's Ryzen 9 6900HS. There is no doubt that Alder Lake's P cores are the most performant for this generation single thread.
There is a mix of good and bad when looking at Handbrake results for CPUs of class 45W and below. The 12900HK is faster than the 12700H in this test, but by only 2 percent, which is relatively insignificant. It can also only match the Ryzen 9 6900HS, even though the 6900HS is a 35W CPU. However, the 12900HK is clearly faster than the M1 Pro, despite our M1 testing using Apple Silicon's version of Handbrake to encode the CPU. The M1 isn't very good at encoding CPU-heavy like this compared to Intel or AMD, which gives the 12900HK a 25% performance advantage.
At 75W, the results are quite similar to Cinebench in that the 12900HK and 12700H give essentially the same results, and both are similarly faster than the 6900HS running at the same wattage limit (4% faster). The high-power configuration sucks the M1 Pro, with the cost of power consumption clearly being much higher than that of Apple's SoC.
The blender is a particularly strong result for Apple Silicon, and here the 12900HK can fit that part, albeit in the higher wattage configuration, 45W versus about 35W.
Again, we didn't see a huge difference between the 12900HK and 12700H for this heavy multithreaded test, and nothing I've seen so far suggests it makes sense to buy a Core i9 for these applications. Even at 75W, which many other Core i9 laptops will be capable of, the gap is negligible for top-of-the-line Intel Alder Lake parts.
As for compiling, our advice would be to buy an Alder Lake processor, as they are clearly faster than competitors and older options.
The 12900HK is 32% faster than the 11980HK in this Chromium compiler and 24% faster than the 6900HS - those are big gains. But again, a very common theme of this review so far, the 12900HK and 12700H are antiquated and make the upgrade to Core i9 a bit pointless.
And then again at 75W, I've seen essentially the same results with the two Alder Lake sections I've tested so far, showing that there's no per-watt performance advantage for the Core i9 at these higher power levels.
In our Matlab testing, the 12900HK and 12700H delivered about the same performance, which is a somewhat surprising result given the short time frame of this workload but here we get similar results.
Aside from the lack of an upgrade from Core i7 to Core i9, this is still a good result as it is 17% faster than the 11980HK and 23% faster than the 6900HS. The M1 Pro is in a rather forgettable position for this app as it has to run in emulation mode which sucks in performance.
In Microsoft Excel we see the same as before, where 12900HK and 12700H are neck and neck. Like Matlab, this benchmark is run in boost and we've seen before that at higher power levels there's really no separation between these two CPUs. So when both hit 100W plus, there is no difference. However, Alder Lake still impresses here, essentially destroying every other part of the rankings thanks to its cache improvements and an upgrade to DDR5 memory.
The PCMark App workload testing the Microsoft Office suite was one of the best for the 12900HK versus the 12700H, and we finally see a separation there.
The Core i9 processor is 9% faster, which gives buyers some reason to get their hands on a Core i9 as that margin is outside our usual margin of calling it a "tie". The 12900HK can also hold leads more handy than the previous 6900HS (15%) and 11980HK (21%).
In 7-Zip compression, we couldn't detect a difference between the 12900HK and 12700H. Intel may have the lead in performance over the M1 Pro, Apple Silicon is faster but the 12900HK is 7% faster, albeit at much higher power since this mostly runs in boost.
But for decompression, the fat P cores can do this task superbly, beating the M1 Pro by a huge 52% margin. There's not much difference between the 12900HK and AMD's new Ryzen 9 6900HS processors, so really all current-generation high-end chips are on par here.
In the Acrobat PDF export, we see the benefit of the higher single-core turbo clock that the 12900HK can over the 12700H, which boosts performance by 4%. Still, that's still not enough to take Apple's lead, the 12900HK is 8% slower than the M1 Pro, which is one of the best Apple results I've tested.
Adobe Photoshop is another light threaded workload, and this is where the 12900HK is quite impressive, scoring the fastest in the Puget Systems benchmarks, 4% faster than the 12700H and 17% faster than the 6900HS. There's a big reason to upgrade from 11th generation silicon or later, as the 12900HK is a sizable 36% faster than the 11980HK.
In FL Studio, the 12900HK is also the fastest CPU I've tested, although it's only 5% faster than the 12700H and the margin between all of these is pretty small for outputting a track at high quality.
Moving on to some mixed workloads now, in Adobe Premiere Pro there is certainly a reason to use the 12900HK profile on the lower end parts, especially with high performance GPUs. The GE76 is about 10% faster than the 12700H model with the same RTX 3080 Ti GPU and this gives a 23% lead over the M1 Pro. The M1 results are extremely impressive given its overall power level and lack of high-performance discrete graphics, but for the absolute best performance of this generation, you should go with something from Intel, specifically. is that the 12900HK plus 3080 Ti configuration is 60% faster at export.
n After Effects, the 12900HK is also the fastest configuration tested and is 7% ahead of the 12700H model using the same GPU. That's one of the better results we've seen so far when comparing the Core i9 and Core i7 models.
Finally, we'll take a look at the photometric performance of Agisoft Metashape, and there's nothing separating the Core i7 and Core i9 systems, which should be adequate for some of our CPU-heavy workloads in this review. . Performance is as fast as you'd expect (and expect) from a high-end system with a powerful CPU and GPU.
Power Scaling, Usage
When looking at CPU power scaling there’s only one interesting observation, and that’s how the 12900HK fares compared to the 12700H in performance per watt. At least with the samples we’ve tested so far, the 12900HK is only a very minor improvement in efficiency, and that efficiency gap tapers off at around 75 to 85 watts.
Given the actual silicon is essentially the same, the main reason for the 12900HK’s improved efficiency is a better bin on those parts, but for this generation it still looks like the 12700H is receiving very good silicon.
In contrast with the 11980HK vs 11800H, there was a more substantial gap in efficiency across the power range – not a huge gap, but indicative of a greater degree of binning.
Without a larger difference between 12700H and 12900HK performance per watt, the best Alder Lake part isn’t quite as efficient as the M1 Pro. We should note that previously we had the M1 Pro at 40W, whereas its actual package power – the same measurement used for Intel and AMD CPUs – is closer to 35W, so we’ve corrected that here.
The M1 Pro is only 6% better at 35W and uses a more advanced 5nm process node, so it's not exactly a great win for Apple in terms of multi-threaded performance - but it's better.
The 12900HK is also slightly more efficient than AMD's new Ryzen 9 6900HS at the same 35W pack power, although the difference is negligible.
Where the difference isn’t so negligible is in single-thread power as measured at the wall.
The M1 Pro clocking up to just 3.2 GHz is ridiculously efficient in comparison to most other processors in this chart, and that’s where it’s insane performance per watt and extended battery life comes from – Apple has a big advantage when tasks are lightly threaded.
In contrast, Intel is going for the win with single-thread performance, and that causes a high level of power consumption, 33W when pushing 5 GHz, which is fairly higher than the 12700H despite just a 6 percent increase in frequency.
We originally intended to run a large part of the gaming performance for the 12900HK, but the results were incredibly boring. Compared to the Core i7-12700H, the 12900HK performed essentially the same on our benchmark suite.
A couple points of margins here and there using the same RTX 3080 Ti Laptop GPU, but on the whole we’d say they perform the same at 1080p gaming, including titles played at low or competitive settings like CS:GO and Rainbow Six Siege.
We have a more detailed breakdown available in our 12700H review, but basically nothing about the 12900HK changes what was seen there. Alder Lake are currently the fastest laptop CPUs for gaming, although the exact margins in comparison to 11th-gen and Ryzen APUs vary, usually up to a 25% gain when CPU limited and much lower when GPU limited. You’ll also need a high-end GPU to take advantage of these differences, and be playing at 1080p.
What We Learned
There’s two ways you can look at the Core i9-12900HK. On one hand, it’s just as impressive as the Core i7-12700H we looked at a few weeks back and packs all the same performance and platform advantages. It’s a fast processor, the fastest on the market for multi-thread, single-thread and gaming – and it has a really robust collection of technologies including DDR5 support and PCIe 4.0.
On the other hand, like with the 11th-gen Core i9 series, the 12900HK feels a bit pointless for most buyers. The performance uplift over the 12700H is only a few percent in most cases, and can actually be the same at higher power levels, pointing to only a limited efficiency improvement.
The biggest gains were seen in single or lightly-threaded productivity apps where the boost up to a 5.0 GHz single-core turbo clock is able to deliver a 10 percent performance gain at best. Of course, if you want the best of the best, the Core i9 is that part, but like with a lot of flagship parts, you’ll end up paying a hefty premium for a very small real-world upgrade.
Purely looking at Intel's line-up, right now I’d definitely be looking at laptops that use Core i7 processors instead of the Core i9-12900HK. Provided the rest of the configuration is the same, you could end up saving a few hundred dollars. But of course, not every laptop will have a Core i7 option. At times the 12900HK might be your only option, and in those situations there isn’t anything wrong with the 12900HK. It very much is a fast laptop processor.
About that M1 Comparison
During Intel's launch presentation, they touted the 12900HK as "the fastest mobile processor ever", showing the slide below comparing their new part to the M1 Max and older generation parts. . Since the M1 Max CPU is the same as the M1 Pro we tested, we can say this slide is partly right and partly wrong.
In this section, where Intel compares the 12900HK to M1 Max at 35W, they show the 12900HK ahead. That can be true depending on the benchmark, but there are more instances where the M1 is either on par or faster at the same power level in our testing.
We’re talking small margins for the most part. The Core i9-12900HK is definitely competitive with Apple in performance per watt, but I wouldn’t go as far to say it’s clearly better.
Where the 12900HK clearly has an edge is in overall performance. At 45W and above, the 12900HK almost always beats the M1 Pro, and when we start pushing up to the 75W range, the Alder Lake Core i9 is a decent amount faster in a lot of heavy CPU workloads. So if you want the absolute fastest workstation laptop for stuff like code compilation, 3D rendering, video editing and so on, the 12900HK will be providing that – especially when equipped with a top end GPU.
But the way in which Intel wins is through increasing power consumption, which is only suitable for certain types of laptop. The MSI GE76 is one of those systems: it’s big, it’s heavy, it’s designed to sit on a desk, and be plugged in most of the time. Intel won’t have the same lead on Apple or AMD in thinner battery-oriented systems; performance will be competitive, but you won’t see 15 or 20% better performance for CPU heavy tasks.
It's obviously bigger and heavier than something like the Ryzen 6000-powered Zephyrus G14, so it's less portable and less likely to be used on battery power. I certainly wouldn't want to use a 3kg laptop on the go. So if I were to sit here and say that the GE76 has worse battery life - and let's be clear, it certainly does - I'm simply comparing apples to oranges. And battery life isn't just based on the CPU but on the entire platform including the GPU, display and other components, so making broad Alder Lake battery life claims is based on a system will be misleading.
To make a fair comparison between a CPU like Core i9-12900HK in terms of battery life with other laptops we need basically the same laptop for testing or alternatively some kind of system the very same. Hopefully that will be possible in the future, but not currently. All we can tell you is that the MacBook Pro has great battery life, the Zephyrus G14 is pretty good, and the GE76 isn't great. Overall, Alder Lake's battery efficiency fair will remain a mystery until we're able to test the mobility-focused systems first.