Alder Lake CPUs: Overclocking and general ADL memory/motherboard discussion

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
@ EarthDog
Are you running it without throttling on 12900K at over 241W and full load in the AVX stress test? This is the only scenario where I see it throttling on the 360 AIO. AIDA64 with CPU+FPU test shows ~18% throttling here and there on single cores while most cores are below 100°C. At the same time, hwinfo64 shows 258W under full load. If I'm right then Johan has the same results on his setup.

@Voodoo Rufus
A custom loop won't be much better because the main problem is with the heat transfer from cores to the block/liquid. If you could keep the liquid much cooler than in the AIO cooler then a custom loop would handle it below the throttling point. If you add a graphics card to the loop then it won't happen.
 

EarthDog

Gulper Nozzle Co-Owner
 
 
 
Joined
Dec 15, 2008
Location
Buckeyes!
My stress test is CPU/FPU/Cache with AVX for Aida. The only board that throttled at stock so far was the valkyrie.
 

Johan45

Benching Team Leader Super Moderator
 
 
 
Joined
Dec 19, 2012
@ EarthDog
Are you running it without throttling on 12900K at over 241W and full load in the AVX stress test? This is the only scenario where I see it throttling on the 360 AIO. AIDA64 with CPU+FPU test shows ~18% throttling here and there on single cores while most cores are below 100°C. At the same time, hwinfo64 shows 258W under full load. If I'm right then Johan has the same results on his setup.

@Voodoo Rufus
A custom loop won't be much better because the main problem is with the heat transfer from cores to the block/liquid. If you could keep the liquid much cooler than in the AIO cooler then a custom loop would handle it below the throttling point. If you add a graphics card to the loop then it won't happen.

Yes, I have had this problem with the Valkyrie, the MSI is very close but taps out running Blender and throttles. I found that setting the voltage manually resolves this issue. The boards are supplying too much IMO the Valkyrie in particular is nearly 1.4V. Dropping it to 1.2-1.25V everything runs stable and the throttling is gone. I can set my OC (5.1 and 4.0) with 1.25V and run cooler than stock.
 

EarthDog

Gulper Nozzle Co-Owner
 
 
 
Joined
Dec 15, 2008
Location
Buckeyes!
Same story here, Johan. This is how all my boards behave. Only the Valkyrie throttles, the rest don't seem to. When overclocking to 5.1/4.1 @ 1.25V I don't have that issue.
 

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
Gigabyte (at least on F5 BIOS) goes up to 1.45V at auto. The same is on Valkyrie. I wasn't checking Gigabyte with F6b BIOS as it was released yesterday. The only thing I noticed is that memory training is a bit better and doesn't require turning off the PC to accept memory timings.

Little update:
12900K on ASRock Z690 Extreme goes up to ~215W max in the CPU+FPU AIDA64 test. Noctua NH-U12A keeps it at about 92C with a single spike to 99C, but there was no throttling reported in hwinfo64 and AIDA64. All P-cores at 4,888GHz. VCore was reported as 1.256-1.305V during the test. This is at auto settings.
 

Zantal

Member
 
Joined
Mar 19, 2009
Hello boys,
Just got back into overclocking since I got my new 12900k 2 weeks ago and so here I am again after years!

So far it's going well but I would like to know if you have the same issues I am currently facing.

Managed to get a stable overclock on all cores at 5.2 with less than 1.39v (4.0 on E-cores)
(5.3 ghz on auto vcore would pin core temps to 100 with 1.55 vcore but still stable so I am settling for 5.2)

I still haven't run any real stability test, but importing textures in unity3D is taxing enough to check for stability issues

The main problem is that I have 2 cores that will reach a max temp of 80C while the others can get to 94+
This is reflected in normal use as well although the difference between the lowest and highest core gets higher the more energy i put into it.

Now, I am used to my old skylake having core to core temp variations of about 5/7 C under load, but 14 or more is kind of worrying me

I am using a noctua D15, it has the proper adapter (although I the skylake one worked just fine) and I reseated it to be sure at least 3 times, nothing changed.

Or maybe I should look at it differently, all the other cores seem to be in line with every review I have read online regarding temps at least, while those 2 cores are unusually cool o_O


Anyway so far this thing is a beast
 

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
This is normal. On my Noctua NH-U12A or even 360 AIO, 1.4V+ causes throttling and some motherboards set up to 1.45V at auto. ~1.3V seems optimal to keep it without throttling and at still acceptable temps. New CPUs simply run hot, no matter if it's AMD or Intel. Various power saving technologies keep them below 100°C but if there wasn't any power limit then all would hit a throttle point.

Noctua NH-D15/S should be more than enough but depends on the motherboard, you may see higher or lower temps so better is to set voltages manually.
If 2 cores run cooler then probably the CPU isn't loaded to the max. For me and on mentioned Noctua NH-U12A, I see about 92-95°C on all cores when the CPU runs at about 1.26-1.30V in stability tests.
 

Zantal

Member
 
Joined
Mar 19, 2009
Noctua NH-D15/S should be more than enough but depends on the motherboard, you may see higher or lower temps so better is to set voltages manually.
If 2 cores run cooler then probably the CPU isn't loaded to the max. For me and on mentioned Noctua NH-U12A, I see about 92-95°C on all cores when the CPU runs at about 1.26-1.30V in stability tests.

The only "stress" test I made was with Cpu-z, hwinfo show 100% core utilization, no temp or power throttling, and the scores I get with it are in line if not higher than other people's results.
This is going to be a mistery.

Maybe I have 2 really good cores while the rest are average.

I will try and get some real bench and stress tests to compare with what you guys have posted here so far to see if those cores are behaving badly
(or run single threaded benchmarks on those cores only)
 

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
In some programs, you can pick single cores. You can also set affinity to specified cores manually in the task manager. I just don't think it will help much as some cores are always better than others and some are already "signed" to boost higher (usually 0/1).
There is also a difference between cores because of their placement and other factors. When you load all cores to the max then the difference between can be 2-10°C. It was typical for most previous generations so I assume that now it's the same.
One more thing can be the cooler's mount. I've noticed that some coolers don't put equal pressure on the whole IHS surface. However, I have no problems with Noctua coolers.
Right now I'm testing 12600K with Noctua NH-P1 in the passive mode and the difference between the coolest and the hottest cores is about 6°C.

To test full load, check AIDA64, Prime95 or any other popular stability test. CPU-Z is pretty bad for everything.

Memory training seems annoying on all Z690 motherboards so far. There are more or fewer problems, but more with DDR5. Gigabyte with DDR5 at any manual settings is retraining RAM every time when I turn it off and on. After that, it is sometimes stable and sometimes not. Like when it passed 2h stability test yesterday, today at the same settings is crashing in 10 mins. On ASRock, with DDR4 I was testing RAM at 4000 XMP and Gear 1. All was fine for a couple of hours. The next day it couldn't enter Windows at the same settings :bang head After clearing CMOS it works again.
 

Zantal

Member
 
Joined
Mar 19, 2009
AIDA64 stable at 5.3 1.45v all core
prime95 is another beast, constant crashing if it starts at 5.3, ok with 5.2 but then quickly throttles down to 4.9 due to heat limit, I think I might get it stable at 5.3 (for that half second anyway) with higher voltage, but my rule is to not exceed 1.45 VID
I'll get it stable at 5.3 for gaming and work mainly, it's difficult I will ever encounter prime95 load types anyway, also want to get a little bit more out of the E cores, aiming for 4.3 as I am currently at 4.2

I don't know if it's an ASUS board issue, but it seems I cannot alter the v/f curve even though the option is present, it will simply not post the moment I change something for the last voltage point.
trying to control voltage with offset is not so straightforward as there seems to be 2 settings you can change, one for default Core/Cache and another for PL2.
PL2 settings seem to be totally ignored, maybe there is something I am missing here

Btw motherboard is an ASUS Z690 prime D4
 

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
If you run AIDA64 with FPU+CPU tests only then it should give similar results to Prime95.
I don't have ASUS Z690 mobo but I see that all brands have problems with BIOS, more or less.
 

Voodoo Rufus

Powder Junkie Moderator
 
 
 
Joined
Sep 20, 2001
Any thoughts on the max ram speed in DDR4 for Gear 1 and 1T on any of these boards?
 

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
I think I mentioned it in some posts. Max Gear 1 on ASRock and MSI is 4000. Biostar has problems at 3600+. Max CR1 depends on the motherboard but with Gear 1 will be also 4000. I wasn't testing ASUS and from Gigabyte I have only DDR5 mobo. I think that PolRoger could make 4000 Gear 1 on his Gigabyte.
I guess that you could push Gear 1 to something like 4100-4133, but there is one problem. At least on ASRock there is no memory ratio for 4100-4133 so the next ratio is for 4200. At 4200 I couldn't make it boot with Gear 1.

I had no time to check high memory clock on Z690 with Gear 2, but on Z590, 3733 Gear 1 was slower than 4600+ Gear 2 and I could push memory to 5000+. Barely anyone is testing RAM at high clock and Gear 2 because most are reading posts/articles of clueless people who without tests repeat that Gear 2 and high clock is bad. On Intel, performance loss is not as significant as on AMD. I don't want to say what is better or worse but it's worth to test high clock with Gear 2 and compare. There are always tests/games that react better to low clock/tight timings and high clock/more relaxed timings.
High clock/Gear 2 is a good option for all who have memory kits based on Hynix A/D or Micron B. In both cases, can run them higher and at lower voltages than Samsung.
 

Johan45

Benching Team Leader Super Moderator
 
 
 
Joined
Dec 19, 2012
Any thoughts on the max ram speed in DDR4 for Gear 1 and 1T on any of these boards?

With an updated BIOS the Biostar Z690-A Valkyrie can run DDR4 4000 CR1 in Gear1. I haven't completed my testing test and highest kit I have will only go to 4400 Gear2 for comparisons.

4000 stability.jpg
AIDA cache 4000.jpg
 

Attachments

  • 4000 stability.jpg
    4000 stability.jpg
    169.3 KB · Views: 135
  • AIDA cache 4000.jpg
    AIDA cache 4000.jpg
    76.9 KB · Views: 135

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
Here is 4600/XMP settings at Gear 2 and initial BIOS. I bet that on the new BIOS it works better. It's good to see it works at 4000 CR1 and Gear 1. I'm still on ASRock and results are about the same, limited by the platform rather than motherboards.

Biostar_Z690AValk_R1.jpg
 

Attachments

  • Biostar_Z690AValk_R1.jpg
    Biostar_Z690AValk_R1.jpg
    222.8 KB · Views: 135

PolRoger

Member
 
Joined
Jul 31, 2005
Any thoughts on the max ram speed in DDR4 for Gear 1 and 1T on any of these boards?

There is a thread running over on OCN Forums covering Z690 and DDR4 motherboards. https://www.overclock.net/threads/official-intel-z690-ddr4-daily-memory-overclock.1794799/

Like Woomack mentioned ~4000/4100/4133 seems to be ~max for Gear 1... Sometimes 1T sometimes 2T. Some of the people in that thread are running rather hefty voltages too. Which may be OK for benching purposes but some of the settings I've seen seem a little high to be running long term daily or 24/7.

2x1R sticks is easier than 4x1R and 2x2R sticks followed last by 4x2R sticks. Most of the people that I've seen are running either 2x1R or 2x2R kits.

Not surprisingly there seems to be a bit of a silicon lottery going on with the chip samples and IMC. How high one can post to BIOS vs. actually booting into Windows with light stability vs. heavy memory stress stability in Windows.

With my low bin 12600K I can run 3900 Gear 1 with good daily stability but my sample seems to be hitting some kind of IMC limit/wall at 4000 Gear 1?

I can boot into Windows with light stability but when attempting to bench something like HyperPi... It will throw out errors. Still a work-in-progress and perhaps a new BIOS might help but it could be that for this sample 4000 G1 is the max and not fully stable??


i5-12600K P49/E39 3900C16 DRAM 1.40v, VDDQ 1.30v, VCCSA 1.25v:

i5-12600K P49E39 3900C16 G1.png
 

Attachments

  • i5-12600K P49E39 3900C16 G1.png
    i5-12600K P49E39 3900C16 G1.png
    460 KB · Views: 126

Voodoo Rufus

Powder Junkie Moderator
 
 
 
Joined
Sep 20, 2001
4000 Gear 1 and 1T was exactly what I was looking for. Can't justify another electronics purchase yet, but a mature DDR4 setup sounds pretty darn good compared to expensive and immature DDR5.
 

Woomack

Benching Team Leader
 
 
 
 
Joined
Jan 2, 2005
The only exception seems Kingston Beast 2x16GB DDR5-5200 that costs less than many, higher series Samsung B, 2x16GB DDR4-3600/4000. I guess that some stores are bumping prices but at least can find it cheaper and it's even available in some stores as almost the only DDR5.

If you get DDR4 then I recommend dual-rank modules as they are 0-7% faster at the same clock. Since Gear 1 is limited to ~4000 then the sweet spot would be 2x16GB Samsung B dual-rank 4000 CL16. However, a good dual-rank Samsung B isn't cheap. On the other hand, if you are planning on keeping the PC for longer then maybe it's worth paying some more as there won't be anything better. Manufacturers already focus on DDR5.

@PolRoger, I don't know what about Gigabyte but on my ASRock, Gear 1 is sometimes acting weird after retraining. I mean it works stable at 4000 Gear 1 but when I try to adjust timings and the motherboard does full retraining then sometimes it has stability problems. Then I have to clear CMOS, set the same again, and count it will work after the next retraining.

My Gigabyte/DDR5 mobo is retraining for a loooooong time. It usually takes 2-3 minutes and I never know if it's going to work or not. I actually gave up until the next BIOS update as each time I set something that passes stability tests, then after restart, it sometimes works or not but usually not and I have to set everything again. The only problem-free settings are XMP1/XMP2 or default/auto at 4800.