Intel i7-12700KF Review – Better than the 5900X?
It’s taken a little while, but I finally have my hands on Intel’s i7-12700K, specifically the 12700KF, thanks to the lovely folks at Cyberpower who sent over an entire system AND both a retail sample 12900K, and this 12700KF. I’ve already tested both the i5 and i9 in a video recently comparing those retail chips with my review sample ones, so do check that out in the cards above, but now it’s time to check out this bad boy, and see how it stacks up against its peers.
Just so everyone is up to speed, this is the i7-12700KF. It’s a 12 core chip, based on 8 performance cores and 4 efficiency cores, meaning it has 20 threads total. It runs at up to 5GHz on a single core with Turbo Boost Max 3.0, although it’s base clocks are actually lower than the i5-12600K at 3.6GHz for the P cores and 2.7GHz for the E cores. By default it’s capped at 190W of Maximum Turbo Power, and can use up to either DDR4-3200 or DDR5-4800 (within it’s warranty limits anyway). The “F” on the end of its name means this is the version without integrated graphics, although the specs are otherwise identical between the K and KF models and when not in use the iGPU shouldn’t be using any power.
Naming wise, it should line up with AMD’s Ryzen 7 5800X, although specs and as you’ll see performance wise it’s more in line with the Ryzen 9 5900X, which is likely why it’s priced neatly between the two of them. So, how does it perform?
Well, starting with the CPU specific benchmarks and Cinebench, in single threaded work, unsurprisingly, it sits right between the i5-12600K and i9-12900K, and squarely at the top of the field, holding a healthy lead over, well, everything else. In multi-threaded, despite it only having 8 high-power performance cores available it still just about manages to beat even AMD’s ‘full fat’ 12 core 5900X. It’s within spitting distance, but it’s still a victory.
In a slightly longer workload though, like the BMW scene rendering in Blender, the 5900X manages to scrape the lead by just a single second. If that’s not trading blows I don’t know what is. Interestingly, the 8 core 5800X had a rough time by comparison, even being beat out by the “10 core” 12600K with its 6 performance cores and 4 efficiency cores.
Interestingly though, in the much longer Blender Gooseberry render, the 5900X extends it’s lead so much that it’s only a hair behind the “16 core” i9. The 5800X catches up too, still a fairly significant portion behind this i7, but now ahead of the 12600K.
If Premiere is more your thing, the 12700KF is clearly a solid choice on performance alone, with a reasonable lead over even the 5900X, and not far behind the i9 too. After Effects has a more sizable lead over everything – save for the 11600K which I still can’t figure out why this keeps running so high. Anyway, the 12700KF and 12900K score closely together, showing the diminishing returns of the higher end chip at least in After Effects. Finally in Photoshop it’s a similar story, with the newer 12th gen chips holding a solid lead over most everything else.
When it comes to gaming, I had a rather strange time with CSGO which I’m putting down to the hybrid architecture causing trouble again – you’ll see why I say that later on but for now this is the stock out-of-the-box experience I got. It’s still not what I’d call ‘bad’ performance, but it’s far from perfect as even the i5 managed almost 100FPS more, only the 11600K scored lower here.
The Cyberpunk results look much better, with the i7 sitting second only to the i9, just how Intel wants it – and importantly those three have a healthy gap to the rest of the field. It’s not absolutely night and day, but 20FPS especially between 120 and 140 FPS isn’t exactly insignificant.
Watchdogs Legion shows the same setup, with the i7 sitting nicely between the i5 and i9, and all three sit comfortably above the rest of the field with a 26FPS lead over the 5900X. Sadly for Intel, the same can’t be said for Microsoft Flight as all three Ryzen chips hold basically the same position the Intel ones did in Cyberpunk, at around 140FPS versus the Intel chips around 120FPS. It is worth mentioning these benchmarks are all at 1080p and generally medium settings, you can expect a lot less of a difference from any of these chips in any of these games at 1440p instead.
Lastly we have Fortnite, which much like CSGO sees the 12700KF running at practically the bottom of the pile. It’s still a very tight grouping so I wouldn’t be too worried, as things like location on the map and what you are doing will vary your actual frame rate more than really any of these chips will.
So, on the face of it, the 12700K is a pretty decent chip both for productivity and gaming, with the catch that as I’ve been saying since launch, if you buy any of these you are paying to be a beta tester. There are bugs, there are issues you’ll run into and it seems like it’s taking a very, very long time to get even a single update out. Still, the raw power is clearly impressive.
Now, most channels would normally leave that there, but if you’ve seen my other Alder Lake reviews you’ll know what’s coming now. Here is how each of the types of cores perform independently of each other! Starting again with the CPU specific tests, you can see in Cinebench R23 single-threaded that the 12700KF actually improves it’s score when just using the performance cores. Not by too much, but a little. Whereas those E cores, yeah they are about half as fast.
In multi-threaded though, that’s where it gets interesting as even with those four E cores disabled, the 12700K is still faster than both the 12600K with all ‘10 cores’ enabled, and the 5800X, and by a very considerable margin. Of course with the 4 E cores seeing as they are the same four you’ll find in the 12600K, but with an ever-so-slightly higher boost clock, they perform very similarly.
Blender also shows some interesting results, as in the BMW scene the i7’s 8 P cores still outperform everything from the 12600K, 10850K, 5800X and down, and of course the E cores.. Well they aren’t exactly fast. In the Gooseberry scene I was surprised just how close the 5800X gets to the 8 P cores from the i7 – I mean 42 seconds per frame is still incredibly substantial but as a percentage there isn’t all that much in it between them.
In Premiere Pro the 12700K’s P cores practically tie with the 5900X, and bring the 5800X a lot closer into contention. The E cores still lag significantly, again matching the i5’s performance. In After Effects, the P cores show a substantial improvement of almost 10% more performance when not impeded by the E cores. That’s impressive, although strangely the E cores on their own actually scored worse than the 12600K’s E cores and I’m not so sure what’s going on there. And in Photoshop we have another pretty strange result, as the 12700K’s P cores quite massively take the lead over everything including the 12900K’s P cores. Again, I’m really not sure what’s going on here, but it does confirm the pattern that in some programs you may be better off disabling those E cores.
Gaming holds a few answers from earlier, specifically in CSGO where the i7’s P cores score significantly higher than the stock result. It’s still not where I would expect it to be, but it’s a significant improvement over stock. That’s the same in Cyberpunk where the 12700KF sees a 10 FPS increase from disabling the E cores. In Watchdogs the results stay the same, even down the the 1% lows, and Microsoft flight actually sees the performance decline just slightly with only the P cores. The E cores are taking the back of the field as usual, although still remarkably playable at least with the RTX 3080 I’m testing with. Lastly in Fortnite you can see the reason I expected some “shenanigans” with the hybrid design before, as with the E cores disabled it jumps to pretty much the top of the chart, albeit still a very tightly packed chart, save for the E core only runs which again were remarkably playable.
So, is the 12700K(F) better than the 5900X? Well, it is in some regards, generally speaking it’s 1080p medium settings gaming performance is stronger, but I’d argue if you are spending around £400 on a CPU you probably aren’t gaming at 1080p so any gap present definitely shrinks as the resolution increases. In CPU heavy work it’s almost always faster, especially thanks to that faster single threaded performance, although I should note I’ve been testing with DDR5 for these Intel CPUs which if you can even find some is a hefty premium over the DDR4 you might already have, and what you need for any of the non-12th gen chips. You can use DDR4 with these though, and if you want to see how that compares do check out my video on that in the cards above too, but it can be a little hit and miss.
When it comes to cost, while it’s easy to just look at the cost of the chips and declare the i7 the victor, even if you exclude the cost of the RAM, the motherboards are a fair bit more expensive especially compared to a reasonable B550 board for the Ryzen CPUs. The Z690 board I’ve been testing in is an Asus Formula board which will set you back something like £600, and even if you go with a more ‘mid range’ DDR4 board like this Z690 STRIX A D4 you are still looking at more like £320, instead of between £120 and £180 for a decent B550 board instead. Factoring that in definitely swings the value proposition, as does the 190W power usage at stock (versus 142W for the 5900X), which means a bigger cooler might be on the list too.
On the whole, the i7 is a pretty decent chip. It offers great single and multi threaded performance, it’s generally good for gaming, with the catches being price, power and bugs. Seeing the P and E core performance on their own is honestly getting me more excited for versions of these chips that don’t have any E cores – 6 and 8 performance core only options can’t come soon enough.
-
TechteamGB Score
