Intel 12th-Gen CPUs

  • Thread starter Deleted member 197115
  • Start date
XMP I will use partial profile with some settings on Auto and detected on boot, XMP II is complete DIMM profile.
Try to raise RAM voltage if it doesn't boot, mine was unstable on default XMP with Z390.
Thanks for the suggestion. I tried a couple of higher voltage levels without success.

My memory was purchased in 2018. Is it possible it will not be compatible with XMP on newer boards?


Screenshot 2021-12-04 215559.png


Looks like it is running at 2133 MHz?

Would I be better off buying newer DDR4 ram?
 
  • Deleted member 197115

Before you go there try to drop frequency, step at a time.
 
I just saw this. It looks like 1.25v should get me to Gear 1 4000, whether it really does or not should be interesting.

Settings as follows allows Gear 1 to 4000 MT easy. Effectively allowing the memory controller at IMC 2000 MHz,
using a Gigabyte Z690 Pro DDR4 motherboard (should be about the same for other Z690 DDR4 boards too).
VCCSA at 1.25v allow post up to 4133
Raising VCCSA to 1.3v allows post to 4200.
Raising VCCSA to 1.35v allows post to 4266.
Pushing VCCSA up to 1.45v no more gains to be had, couldn't make it to 4300, no post.


What I read was that just using XMP profiles wouldn't work and that you have to manually set the voltages. But mode 1 at 4000Mhz sounds good to me :)
 
Last edited:
Merry Christmas to all.

Well I finally got my new PC up and running, so I thought I'd share some things for anyone interested.
12900K, Asus Tuf Z690 Plus Wifi D4, BeQuiet Silent Loop2 360 AIO, Asus Tuf OC 3080Ti, 32G GSkill DDR4 Ram, Silverstone SX1000 PS (SFX-L) and I went for the LianLi O11 Dynamic Mini case (currently 2x 140mm fans intaking from the bottom of the case, while the AIO rad is mounted on the top exhausting up/out. No other case fans yet)

Here's the performance from CineBench R23, 10 minute stress test, with all panels/filters installed on the case.
- score 27,150. CPU temps between 75-78C max. No throttling or other issues seen after running the 10 min test several times.
(XTU benchmark was 9250)

Other misc comments in case interested:
- I used Intel XTU to undervolt the CPU a bit. 10 min Cinebench was hitting 82-85C (241W power draw), but the fans were getting quite loud. I under volted with a -0.055 offset and the Cinebench score remained the same, temps dropped to 75-78, fans of course ramp up but are more reasonable, and everything still stable with no throttling or other issues seen. The test results in max 4.93GHz, 1.273 max V, and power draw 212-214W.
- I use Thermal Grizzly Conductonaut for thermal paste
- No special fan curves or anything yet. All is completely stock and I'll tweak a bit over time.
- Note that with this case/MB/Ram/Cooler combination, there is not enough clearance at the top of the case to mount the rad without a few mods. First mod was that I cut off one corner of one of the 120mm fans (just where the mount is, so yes, that fan is mounted with 3 screws, not 4) otherwise it was interfering with the heatsink on the MB. Second issue, the middle fan interferes with the ram. It's very close, but I changed out the middle fan on the rad for a low profile one (NF-A12x15). I still had to shave down about 1mm of the fan edge at the exact point where the ram is, just to keep a very small clearance.
- I didn't really comment, but the GPU really has no issue staying at decent temp. Under Heaven testing, it doesn't get above 65C and fans stay at 84%. I'm sure things will evolve when gaming, but so far it seems reasonable to me.
- My original plan was to re-use my Noctua NH-D15 instead of the new AIO. I changed my mind only because I want to keep my old PC still operational, so I was gonna have to get a new cooler anyway and it seems to be working very well cooling my i7-8700k (OC5.0, delided)

No other testing so far other than re-running the Cinebench 10 min test a few times in a row for good measure. All temps written above are after the last test so the system should have been quite saturated and at the worst possible temps. Seems good to me (but I have no idea what I'm talking about....). I didn't explore any real overclocking, and probably won't (other than the under-volt which did make a big difference at least in the fan noise, and also the temps).

Next steps are pretty simple.
- clean up the wiring a bit. The stock Silverstone cables aren't very pretty, and I've been experimenting with some cable sleeving DIY. Might work on that another day.
- I *might* install another fan on the case rear, likely as intake, just to feed some additional cool air to the rad mounted at the top of the case. Have to check the fan curves and see if I can tweak anything a bit.
- Probably spend a bit of time putting some custom lighting and other 'racing inspired' graphics/mods inside the case.
- Install various other software and drivers. Start to reconfigure a few things like audio etc.
- Get racing!!!
 
Now that the PC is at home and the surprise part is over, I had a bit more time to tune it. For a 12600k, this is about as high as it's gonna get. When I have more time, I'll get cache at 44 and ecores at 42 stable but that's just for fun. They won't add anything notable to performance. I used my old 'crap' kit on there thus the meh mem. Maybe I'll swap over my good kit which can do 15/4100+

Performance:
unknown.png


Stability tested:
unknown.png
 
Last edited:
Alderlake 12600k testing various configurations. Everything is overclocked and stability tested with OCCT Large/Avx2 Extreme. For Shadow of the Tomb Raider, pay attention to the "CPU Game" average, not the main score. For FF Endwalker, Overall score is fine. HT = Hyperthreading Pcores/Ecores/Cache. Where ecores = disabled: Pcores/Cache.

Config 1: HT ON. Ecores ON. 53/41/44. 16/4200 SR mem
unknown.png

unknown.png


Config 2: HT OFF. Ecores ON. 53/41/44. 16/4200 SR mem
unknown.png

unknown.png


Config 3: HT ON. Ecores OFF. 53/50. 16/4200 SR mem
unknown.png

unknown.png


Config 4: HT OFF. Ecores OFF. 53/50. 16/4200 SR mem
unknown.png

unknown.png


VS 9900k HT OFF. 52/48. 16/4400 SR mem *look at CPU Game Average to compare*
unknown.png
 
This is interesting.
IBM is actually creating larger transistors, but stacking them vertically in a new way that is more efficient so they are achieving a solid density and still dropping power consumption for their 2nm technology.

TSMC OTOH claims that that that have achieved a "real" 1nm but it appears to be using exotic materials.

I don't expect to see either of these soon, but IBM's 2nm tech seems more doable with existing technology. TSMC's tech working with MIT appears a bit more like a theoretical breakthrough without a good means of manufacturing.

 
Last edited:
Alderlake 12600k testing various configurations. Everything is overclocked and stability tested with OCCT Large/Avx2 Extreme. For Shadow of the Tomb Raider, pay attention to the "CPU Game" average, not the main score. For FF Endwalker, Overall score is fine. HT = Hyperthreading Pcores/Ecores/Cache. Where ecores = disabled: Pcores/Cache.

Config 1: HT ON. Ecores ON. 53/41/44. 16/4200 SR mem
unknown.png

unknown.png


Config 2: HT OFF. Ecores ON. 53/41/44. 16/4200 SR mem
unknown.png

unknown.png


Config 3: HT ON. Ecores OFF. 53/50. 16/4200 SR mem
unknown.png

unknown.png


Config 4: HT OFF. Ecores OFF. 53/50. 16/4200 SR mem
unknown.png

unknown.png


VS 9900k HT OFF. 52/48. 16/4400 SR mem *look at CPU Game Average to compare*
unknown.png
Dayum!!
Could you tell us your voltages? Especially getting the RAM running at that speeds would be very interesting!
 
@RasmusP

This where I ended up with this. I don't think I can do much more for a daily setup tbh:

unknown.png


Voltages are adaptive with 1.40 bios set with llc4. SA = 1.27v DRAM = 1.54v SA and IA Cep enabled. Avx2 guardband 100 and some sub voltages which are very chip specific like L2, pll's, term, etc
 
@RasmusP

This where I ended up with this. I don't think I can do much more for a daily setup tbh:

unknown.png


Voltages are adaptive with 1.40 bios set with llc4. SA = 1.27v DRAM = 1.54v SA and IA Cep enabled. Avx2 guardband 100 and some sub voltages which are very chip specific like L2, pll's, term, etc
Cool. How much performance in sim titles did you gain?
 
Cool. How much performance in sim titles did you gain?

The pc is for my kid who loves minecraft. I just tuned it for fun but it paid off immensely in minecraft. Just using the igpu, he went from 50 to 80fps in a spot I used for testing performance jumps. Saved us from having to hunt down a gpu!

In the meantime, a buddy who does reviews for a RU publication did some testing with ACC:

IkpZoMiJZJ.jpg
 
The pc is for my kid who loves minecraft. I just tuned it for fun but it paid off immensely in minecraft. Just using the igpu, he went from 50 to 80fps in a spot I used for testing performance jumps. Saved us from having to hunt down a gpu!

In the meantime, a buddy who does reviews for a RU publication did some testing with ACC:

IkpZoMiJZJ.jpg
Wow. I need to study what you have done and try to get max performance. I also enjoy the challenge of getting the most out of my systems.
 
went from 50 to 80fps in a spot I used for testing performance jumps. Saved us from having to hunt down a gpu!
Probably lots of stuff happening at that spot and everything running in one single cpu thread? :D

Awesome performance boost!

Thanks for sharing your voltages!
 
Which Z690 motherboards are easier or harder to overclock or undervolt?

There were complaints last November about Gigabyte BIOS not ready for prime time,
while ASUS’ AI overclocking seems well regarded,
but ASUS ROG Strix Z690-A or ASUS PRIME Z690-A VRM should suffice for 12600k but not 12900k
while MSI MPG Z690 Carbon should suffice for all,
but MSI gets dinged for nonstandard BIOS settings terminology
 
Last edited:
Which Z690 motherboards are easier or harder to overclock or undervolt?

There were complaints last November about Gigabyte BIOS not ready for prime time,
while ASUS’ AI overclocking seems well regarded,
but ASUS ROG Strix Z690-A or ASUS PRIME Z690-A VRM should suffice for 12600k but not 12900k
while MSI MPG Z690 Carbon should suffice for all,
but MSI gets dinged for nonstandard BIOS settings terminology

Asus or MSI z690 VRM is fine.

MSI uses actual terms to spec. It's asus who likes to make them up. Yeah GB is a joke.

Don't rely on auto/ai overclock/undervolt. Do it in bios yourself.
 
Last edited:
Which Z690 motherboards are easier or harder to overclock or undervolt?

There were complaints last November about Gigabyte BIOS not ready for prime time,
while ASUS’ AI overclocking seems well regarded,
but ASUS ROG Strix Z690-A or ASUS PRIME Z690-A VRM should suffice for 12600k but not 12900k
while MSI MPG Z690 Carbon should suffice for all,
but MSI gets dinged for nonstandard BIOS settings terminology
Is it worth overclocking for sim racing? How much fps improvement would you get from overclocking a 12700k? I have assumed it was minimum and have not tried it on my 12700k.
 
Is it worth overclocking for sim racing? How much fps improvement would you get from overclocking a 12700k? I have assumed it was minimum and have not tried it on my 12700k.
Simply take your fps and scale it with frequency.
85 fps @ 4.5 GHz
5 GHz => (85/4.5)*5 = 94.44 fps

The issue is that the newer gens boost very high on default and overclocking can be difficult since you can't get the single cores to boost much higher and you also can't get that boost clock on all cores stable that easily.
 
Last edited:
Simply take your fps and scale it with frequency.
85 fps @ 4.5 GHz
5 GHz => (85/4.5)*5 = 94.44 fps

The issue is that the newer gens boost very high on default and overclocking can be difficult since you can't get the single cores to boost much higher and you also can't get that boost clock on all cores stable that easily.
I thought sim games like ACC were more limited by the GPU and CPU overclocking didn’t do much.
 

Latest News

How long have you been simracing

  • < 1 year

    Votes: 351 15.6%
  • < 2 years

    Votes: 244 10.8%
  • < 3 years

    Votes: 241 10.7%
  • < 4 years

    Votes: 177 7.9%
  • < 5 years

    Votes: 299 13.3%
  • < 10 years

    Votes: 258 11.5%
  • < 15 years

    Votes: 165 7.3%
  • < 20 years

    Votes: 125 5.6%
  • < 25 years

    Votes: 99 4.4%
  • Ok, I am a dinosaur

    Votes: 291 12.9%
Back
Top