Core i9 11900K and Core i5 11600K: Intel Adaptive Boost analysis

One of the more interesting features of the Core i9 11900K is a feature that was announced after the chip was shipped to reviewers, Intel Adaptive Boost or ABT. The idea is to offer a frequency bump to all of the cores that wouldn't have already been boosted by technologies like Thermal Velocity Boost and Turbo Boost Max 3.0, which target the best-performing cores. That helps in situations where many cores are loaded, such as content creation workloads, but is less impactful in scenarios like games where the performance of a smaller number of cores are the limiting factor.

We're not talking a massive overclock here - the boost is up to +300MHz on two cores, +200MHz on two more and +100MHz on the final two in the illustration that Intel provided - but it does mean that the Core i9 11900K should be able to run at an all-core frequency of 5.1GHz if power and temperature numbers are within a healthy range. We wanted to see if that made a noticeable difference to gaming or content creation performance, so we performed a quick trio of game tests and our content creation benchmarks. We also ran our content creation workloads with multi-core enhancement disabled, to see how much of a difference staying within Intel's power limits made to our results as well.

So in content creation, we see a nearly 19 per cent performance advantage for the 11900K by enabling MCE (which is essentially asking the motherboard to ignore Intel's turbo limit guidelines). There's a further four per cent boost in Cinebench R20 and HEVC by turning on ABT, which also results in the highest power usage we've ever seen: 390W. (Meanwhile, the 11600K sees nearly no advantage from enabling MCE in Cinebench, but does gain 14 per cent in the Handbrake h.264 encode.) If your power supply and cooling solutions are a little bit stronger than you really need, then this might be a good way of exchanging some of that headroom for a small performance advantage.

Game benchmarks next, and here we're reprising the titles from our RAM tests plus my favourite go-to, CS:GO. With ABT enabled, we see a seven per cent boost to performance and an eight per cent improvement to the worst one per cent frame-times. That's not bad for a simple BIOS switch. Especially if you're trying to keep your frame-rate in the sweet spot for G-Sync or FreeSync on your monitor, but even if you just want minimal input lag at frame-rates much higher than your monitor's refresh rate, an extra 20fps is always welcome.

Next, Ashes is entirely CPU-bound, even at 4K, and uses as many cores as it can, so it's a natural candidate for ABT to make a noticeable difference. Indeed, we do see a two per cent increase when the switch is flipped, even though that's only good for an extra frame per second. For context though, that's twice the increase that we saw when swapping from 3066MHz to 3600MHz RAM, so we'll count that as a repeatably measurable difference.

Far Cry 5 is basically the opposite of Ashes, heavily relying on one core, so it's not surprising to see a far smaller impact here of only one per cent or about 3fps. Still, we saw similar margins through repeated runs, so it's more than run-to-run variance impacting our results here.

Crysis 3 is last. This game famously doesn't really care about memory speed, and so it goes with Adaptive Boost too. The game is just fractionally faster with ABT engaged, although the worst one per cent scores do rise by 2.5 per cent. Again, not earth-shattering stuff, but it is at least measurable - and perhaps this is something that can be optimised over time.

So there we have it - measurable performance uplifts, but not particularly impactful ones. I'm curious to see if we've missed a trick here by testing with MCE enabled, as perhaps we'd see a more noticeable performance increase when the processor is operating within its stock limits. There was also a BIOS update that hit after we completed our 11900K testing, which uses the same microcode as 610 but may still alter how ABT operates. We'll no doubt come back to this feature in future if we do spot more potential for it, as the thinking behind it does seem sound and we expected slightly better performance from enabling it.

One final note: given that the slide states that Adaptive Boost isn't considered overclocking, this is a nice way to unlock a small amount of extra performance without falling foul of your warranty. Manual overclocking can be a bit nerve-wracking in the early going, so a friendly-looking BIOS switch that says "give me more performance" is nice to have. It's also a necessary value-add for potential 11900K owners - who are otherwise facing a tough time justifying the Core i9 over a cheaper Core i7 that has the same core count and only slightly slower clock speeds. For their sake, I hope other outlets - or future BIOS updates - show bigger improvements, as right now I'd be leaning towards the Core i7 if I was shopping myself.

Regardless, I hope we see Adaptive Boost available on the full lineup of Intel processors in future; more mainstream processors could certainly benefit from the extra performance.

Intel Core i9 11900K and Core i5 11600K analysis

• Introduction, hardware breakdown, test system
• Gaming benchmarks: Flight Simulator 2020, Hitman 3, Total War Three Kingdoms
• Gaming benchmarks: CS:GO, Rainbow Six Siege, Black Ops Cold War
• Gaming benchmarks: Cyberpunk 2077, Far Cry 5, Crysis 3
• Gaming benchmarks: Memory bandwidth analysis
• Gaming benchmarks: Intel Adaptive Boost analysis [This Page]
• Intel Core i9 11900K and Core i5 11600K: the Digital Foundry verdict