Idle power consumption
It’s hard to generalize because whether it’s Ryzen or Core, one little cough in the system and the wattage takes a dive. However, one can also observe situations with around 20 watts (AMD) and a bit under 30 watts (Intel) of package power, whereby the Z590 chipset doesn’t remain modest either in my opinion. The MSI MEG X570 Godlike is around 30 to 42 watts, the MSI MEG Z590 is around 50 watts+ when under full load. At idle, it hardly gets much less with either, especially when PCIe 4.0 is used.
Power consumption in applications
Let’s first take a look at Cinebench, which could also stand for similar applications like Blender or Luxrender in Multi-Core. AMD’s Ryzen 9s are no slouches, but they also deliver a decent performance. Only the Core i9-11900K doesn’t want to fit into the scheme, because it is cooled down well and hunts through the amps that the setting offers. ABT in particular is a nice cost driver there. Clock or not, it would only have to correlate properly with the recorded performance in the performance result, then nobody would complain. But that is precisely not the case.
In single-core, Intel’s Core i9 and Core i7 are well above a Ryzen 9 5950X, which tops the benchmark.
This is even more true for the multi-core benchmark, which illustrates Intel’s problem even better. Tap open, running. By the way, I also tested again with Chiller and Unlimited – almost 390 watts are possible with ABT (which is complete nonsense). The funny thing is that at some point nothing scales anymore that you put more energy into it.
The worst case scenario is Prime95 with or without AVX. It really is possible with proper cooling to beat a Core i9-11900K up to 400 watts, but those are things no one needs. Except for the LN2 faction, but thankfully they are in the minority. I don’t need 7 GHz for the gallery and more than 1.5 volts on the CPU are also very unsuitable for everyday use. That’s why I’m saving those Prime95 and AIDA FPU numbers for another article. Nobody plays it anyway. Talking about games….
Power consumption in games
The following charts once again summarize all the measurements I recorded during the games. We can therefore also compare very nicely where which CPU is working with which result and how efficiently (or not). However, this is exactly where the Intel CPUs almost always have a strong disadvantage, because the power consumption goes up exorbitantly especially in the border area with full CPU load. If you compare the Core i9-11900K with an only marginally slower Ryzen 7 5800X, then there aren’t just worlds, but small galaxies between them.
And if you think that’s because of my BIOS setting with the 288 watts for P1 and P2, you’re sadly mistaken. In fact, setting the CPU to the default value makes it a tad slower in the full load scenario, but it doesn’t make it more efficient. Only a clock and voltage reduction would help there and it is certainly a usable topic for a follow up. Especially Intel’s ABT you notice first at the socket and then (maybe) in the applications. This in turn makes the Core i9-10900K much more interesting, which does without such gimmicks, has two cores more and is hardly slower in the end, but usually much more frugal. Intel’s quest for the performance crown in gaming seems almost grotesque in some situations, regardless of whether a Radeon or GeForce is in the computer. And from my point of view, they missed that target as well.
By the way, it is also worth taking a look at the combined CPU and CPU display. Again, it’s nice to see how little power consumption some CPU can push a fast graphics card to the limits and how long (in contrast) some bars get for nothing. It’ll be crow-bar style and almost bulldozer feel. Ok, I want to be fair, because here at least the overall performance is right and only the plug glows silently in the evening wind. Pure stream romance.
Shadow of the Tomb Raider
Watch Dogs Legion
Far Cry New Dawn
Power consumption as curves: all CPUs, all games and resolutions