• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Intel Core 12th Gen Alder Lake Preview

Well, it looks like Intel's leaks were running on windows 11 before the AMD patch, which reduces performance for AMD by 15%, and AMD is releasing the V-cache chips first quarter of 2022 which they claim adds 15% gaming performance, so add those two together and with the Windows 11 AMD patch, the V-cache chips may be deleting Intel's gaming gains

This is only of concern for those with a 3090 :) And really not even them.

I mean seriously, RKL and Zen 3 were tied at 1440P with a 3080 here at TPU, and at 1080P the 10900K came out on top. Tom's used a 3090 and at 1440p it's basically a tie for anything from OC (or PBO) 5600X/11600K to the #1 slot, and humorously the #2 slot is a OC Comet Lake 10900K. For all practical purposes, the CPU doesn't make any difference now for games with a high end GPU. It was like that for a long time before 30x0 and rx 6xxx came out, it's like that again with RKL and Zen 3. GPUs are maxed out.

I don't think the top end performance of these is going to change until new GPUs come out. You'll probably have to find some other category to cheer about.


1635368772793.png


1635369077544.png
 
Great, so the AL chips are all 241W?



This is gunna be greaaaaat
 
Great, so the AL chips are all 241W?



This is gunna be greaaaaat

Dude that is just the K-series PL2 with PL1 at 125W. You can easily run these at 125W PL1 and unlocked PL2 with a $20 150W rated air cooler, have done it on my 10850K. It's not until you extend Tau and bump up PL1 that you need more cooling. 241W max spike power draw is not bad. These are my 'I want it to be stable and quiet' settings on my 10850K with a cheap Z490M Prime board - 205W PL1 and 262W PL2 with a 28s Tau. No crashes ever with this, so cheap boards can handle it :

1635370475787.png
 
AnandTech only tests with JEDEC rated memory kit, they don't even do XMP. That means when they test a Zen 3 or Rocket Lake they use DDR4-3200 CL20. He's going to stick with it for AL. That is what he's throwing up.

I think it's interesting mostly in that it tells you what a typical OEM like HP or Dell will put in the box. Many Zen 3 / RKL / CML actually ship with crap like DDR4-2666, and those OEMs are 85% of the market, so it isn't wrong really. However IMO it makes his tests somewhat irrelevant for DIY, maybe ok as a baseline but more info needed.
I know, and I agree with that as a testing methodology (though I also think that warrants further memory scaling tests, separate from or integrated into the CPU reviews). I was commenting more on the low speeds, as well as the rather shocking fact that you actually need a 2-slot motherboard to even hit what I would consider baseline JEDEC speeds (yes, I know there are DDR5 specs down to 3200, but anything below 4800 isn't worth much imo - and there are JEDEC specs up to 6400 and provisional ones up to 8400 after all). That 2dps boards can't hit 4800 with only one single rank DIMM populating each channel is really surprising to me.
 
You can easily run these at 125W PL1 and unlocked PL2 with a $20 150W rated air cooler
Define 150W cooler.
My cooler is rated for 150W and the CPU goes to 95+C at 125W package power.

I would NOT want to run Alder Lake on one of those "150W" coolers.

Seriously, taking advantage of a bug to make their performance look better! That's low even for Intel.
As if that is really, definitely, certainly not going to turn into a shitstorm on YouTube a week from now...
 
Lol, intel marketing department is so sneaky to leave out the 5950x in relation of content creation and such. They know the Ryzen's will whoop intel's ass in relation of anything that requires serious horsepower.

Other then that as i suspected the PL2 status of 240W and proberly beyond takes quite more power for only 10% performance increase over a Ryzen. The 5950x or so peaks at 190W in turbo mode.
 
Might be me time to finally upgrade
 
Define 150W cooler.
My cooler is rated for 150W and the CPU goes to 95+C at 125W package power.

I would NOT want to run Alder Lake on one of those "150W" coolers.


As if that is really, definitely, certainly not going to turn into a shitstorm on YouTube a week from now...

I used a Thermaltake Contact Silent 12 150W cooler. At stock it was perfectly fine. I find it hard to believe your 4690K if not power unlocked is stressing such a cooler. Sounds like a bad or very old paste job (it needs to be re-pasted every 3 years max).
 
When someone writes something like this: "We highly recommend you wait until November 4th for independent performance reviews of 12th Gen Core desktop processors before making up your mind on whether to buy or pre-order one. ", you know this product will fail.

I am VERY interested to see gaming performance while running streaming software on a 9th, 10th, and 11th gen CPU and then again with 12th gen and see the results. This could effectively kill dual streaming setups and you wont have to worry about the CPU being a limiting factor because of the P & E cores.


This was already possible with top tier Ryzen CPUs.
 
Interested to see gaming performance on various models and a bit more of a dive into the E cores which I find fascinating.
 
12900k - 8P cores and 8E cores
5950x - 16E cores
Well there are still some types of workloads, especially with full 32-thread multi-threading, where 5950X will perform better. But 12600K with 12600KF wipes out 5600X completely being as fast as 5800X or even slightly faster. Yet being cheaper than even 5600X. That will be new mainstream. Barely even most gamers needs more than 12600K at the moment.

Actually 190W PL1 for 12700K is not bad I guess.
That is PL2. PL1 is 125 W.
Define 150W cooler.
My cooler is rated for 150W and the CPU goes to 95+C at 125W package power.

I would NOT want to run Alder Lake on one of those "150W" coolers.
I'm sure all mother boards will offer a way to switch on PL1 mode. You won't lose any performance in games, as game do not usually run PL2 mode for more than a couple of seconds.
 
"With roughly a quarter of the die-area of a Golden Cove unshackled for power and at the right clock-speed, Gracemont is able to closely trail Skylake in IPC."
Sounds like now they're making considerable progress though comparison between Gracemont and Skylake needs additional data like transistor count, clock speed, etc.
 
Yet being cheaper than even 5600X
thats why every one is waiting for Alder, coz AMD gonna drop prices on 5600 and 5800 if new Intel cpu will become a success, and we already got ton of DDR4 and amd AM4 mobos around, so having 5600X and 5800X more available is pretty neat.
 
We definetly need a 12900 vs 12700 CPU comparison with the later slightly oc to match the 12900 freqs.
My guess is that the 12700KF CPU would be the best perf/price CPU of this generation and the scalpers heaven, if Intel doesn't provide.
Looking forward for both reviews.
 
Seriously, taking advantage of a bug to make their performance look better! That's low even for Intel.

Are you seriously expecting Intel did this on porpuse? Thats an issue to be fixed with AMD and Microsoft, those tests were made Oct 1st. Even intel could improve some stuff with BIOS and stuff from there to now.
 
I dont like it’s high power draw, but it will be the new performance king and AMD has no answer until next year’s fall?
replay to what? most computers are not cpu bottlenecked this is not really any different... the ps4 the most used gaming stsyem right now is 8 cores 8 thradsat 1.6.... cpu in games or most things is not really a issue... a highe end phone is 3x this so unless u are the 2% of highe end users this is a usless cpu on EVERYCASSE use
 
thats why every one is waiting for Alder, coz AMD gonna drop prices on 5600 and 5800 if new Intel cpu will become a success, and we already got ton of DDR4 and amd AM4 mobos around, so having 5600X and 5800X more available is pretty neat.
You see, RKL and CML are much cheaper yet not worse in games. AMD might lower prices, but just a bit. If they could, they would do that already.
 
yet not worse in games
GPU bottlenecks game performance, plus u no longer can buy a highend gpu, that does not costs like an enterprise level piece of hardware.

would do that already.
coz amd still got a good multicore performance, and Alder gonna change it (I wish, coz I already got cool B550 mb and dont mind change 3600 to something better)
 
Some tests suggested the early Win11 build with L3 bug has >10% impact on some games like CS:GO

Should we just take the Intel slide and, modify it with 10%?

Like this?

intel.jpg
 
Last edited:
Some tests suggested the early Win11 build with L3 bug has >10% impact on some games like CS:GO

Should we just take the Intel slide and, modify it with 10%?

Like this?

<snip>

I bet the real difference will be less than 5% on 99% of games.

But then, the real difference between a Comet Lake 10900K and a 5950X is less than 5% on 99% of games. And it's not always in Zen 3's favor.

If you have any high end CPU from Comet Lake, Zen 3, or Rocket Lake you aren't likely to see any big jump in gaming FPS using current GPUs at reasonable resolutions i.e. >=1080P with >= high detail.

Que 720P low detail tests.
 
If intel would just do a 2c 8c desktop that's a unlocked K model for the 2022 models updates they'd have a performance value hit. It would be efficient overall quite parallel relative to cost and the IPC big cores would be a bit like the GHz race in early CPU days yet with power sipping efficiency cores backing them up. Overall the chip would draw about 110w I estimate based on the other TDP's in relation to dropping a few P cores and E cores. That's damn reasonable the max turbo would probably be roughly 100MHz below the 126000K model. It looks like the E cores are overclockable and can be adjusted separate from the P cores as well. I think a unlocked 2c 8c would rip on value for dollar and overall sip power nicely in general.
 
AnandTech only tests with JEDEC rated memory kit, they don't even do XMP. That means when they test a Zen 3 or Rocket Lake they use DDR4-3200 CL20. He's going to stick with it for AL. That is what he's throwing up.

I think it's interesting mostly in that it tells you what a typical OEM like HP or Dell will put in the box. Many Zen 3 / RKL / CML actually ship with crap like DDR4-2666, and those OEMs are 85% of the market, so it isn't wrong really. However IMO it makes his tests somewhat irrelevant for DIY, maybe ok as a baseline but more info needed.
There is a reason to read multiple sites, Anand gives worst case, GN gives best, and TPU tends to give about avg.


If intel wants any credibility they have to fire Ryan Shrout... benchmarking on a build they know hurts AMD, giving them lower than recommended memory speeds.
Claiming a win when they are barely on par, and releasing the slide deck when AMD is showing off its Quarterly performance.
 
Back
Top