Alder Lake CPUs common discussion

[PolRoger]

My friend just picked up 12700K well and it's overheating like crazy on his AIO... here is his 12700K at 240W-280W at 4:9 all core, cannot get 5.0 stable at the moment, here is his cinebench @ 4.9 95C temps:

Seems the 12700Ks are much harder to push? W1zz's was 5.0 at 1.4V ~ seeing 3 other reviews of 12700K deeming it 'not worth' to overclock. Could just be the lottery/bioses? Seems weird for 2 more cores to have such a difference (altho TPU 12600K also didn't really OC).

@PolRoger -- what motherboard are you running?

I was testing my chip at 50x with Prime95 non-AVX 6C/6T and E cores disabled using a thin 240 rad. Stress was running fine at first with temps in the low to mid 70's. I left the room for a while and when I came back later to check it had completely heat soaked the radiator and temps were in the upper high 90's and some cores had bumped up to 100 c and thermal throttling had been triggered. It was still running Prime with no errors though. I shut it off!

 

[Psychoholic]

I have my 12900k limited to 200w package power, it doesnt seem to drop the score much.
Actually, i feel like the sweet spot is around 180w package power.

180 Package Power maxes around 70c and doesnt drop scores much more than this.

I havnt even attempted to OC it because its so damn fast anyway i see no need.

 

[Deleted member 24505]

I intend to get a 12700k soon, probably this week, i think i will just run it stock tbh till there's a few more on here with them fiddling. I have no idea on board yet though apart from z690 and DDR4.

 

[freeagent]

I think it's mostly placebo but it for sure feels faster .

5.4 failed unfortunately - died after some time with the LinPack... back down to 5.3 now and going to run it for a few days before moving on.


*snip*

Havent touched the E cores yet but in 3900x cinebench mt territory.

Man, that thing is a savage. Very nice.

 

[Caring1]

I was testing my chip at 50x with Prime95 non-AVX 6C/6T and E cores disabled using a thin 240 rad. Stress was running fine at first with temps in the low to mid 70's. I left the room for a while and when I came back later to check it had completely heat soaked the radiator and temps were in the upper high 90's and some cores had bumped up to 100 c and thermal throttling had been triggered. It was still running Prime with no errors though. I shut it off!

Would a decent tower cooler with airflow over the VRMs make a difference?

 

[PolRoger]

Would a decent tower cooler with airflow over the VRMs make a difference?

I'm running open air bench style... I prefer to run thick 420 rads on my main rigs... Currently 5950/5900X. From what I've seen the 12600K runs fine @ stock settings (all core) which drops to 45x on my setup. With a decent tower cooler and or a thin 240 type rad I was ok @48x all core. If you want to run Alder Lake @50x-52x(+) all core long term with high loads you will need very good top notch cooling.

 

[AlwaysHope]

FWIW only memtest86 can test your entire memory. Its free version is enough for absolute most average people including OC/tech enthusiasts. Highly recommended. The application is signed so there's no need to disable EFI secure boot.

View attachment 224142

Windows built-in memtest can do it too but it's quite simplistic and not as thorough as memtest86 which can find errors not detectable by Windows memory checker.

I agree with you on Memtest86 but your aware that Windows has 3 levels of its memory checker & cache can be disabled?
Don't think MS are going to code poorly for an app if its flagship OS is relying on it.

 

[phanbuey]

Alder lake REALLY scales with cache & RAM -- ring frequency, in the games im testing, makes a bigger fps difference than core clock - SOTTR, BL3 and HZD benchmarks, and CP2077 and Outer worlds playtime.

Here is the pass 1 tuned result -
ram 4x8gb 4000MHz Gear 1 (2000% kuhru ram test stable) - 17 18 18 37 | TRRDS/TRRDL/TFAW = 4/5/20 | TRFC= 400(200ns) - no RTL/IOL tuning yet.
Cpu - P cores 5.3ghz at 1.30-1.32, E cores at stock
Ring - 44x

The biggest jumps were all from RAM and RING (as is usually the case in SOTTR - but I am also noticing it more in other games (Outer Worlds etc.)) - markedly smoother with ring and cache OC, minimal difference in core clock.

3080 is undervolted @ 1815mhz -- not bad for a "6" core. Biggest jump in Timespy score was also ram+cache driven. Basically miniscule difference from 5.0 to 5.3 core (except the massive echub). Huge double digit % gains on cache and ram to 4000 Gear1. Platform seems starved of memory/cache with heaps of clock/ IPC in the tank.

 

[RandallFlagg]

Alder lake REALLY scales with cache & RAM -- ring frequency, in the games im testing, makes a bigger fps difference than core clock - SOTTR, BL3 and HZD benchmarks, and CP2077 and Outer worlds playtime.

Here is the pass 1 tuned result -
ram 4x8gb 4000MHz Gear 1 (2000% kuhru ram test stable) - 17 18 18 37 | TRRDS/TRRDL/TFAW = 4/5/20 | TRFC= 400(200ns) - no RTL/IOL tuning yet.
Cpu - P cores 5.3ghz at 1.30-1.32, E cores at stock
Ring - 44x

View attachment 224755

The biggest jumps were all from RAM and RING (as is usually the case in SOTTR - but I am also noticing it more in other games (Outer Worlds etc.)) - markedly smoother with ring and cache OC, minimal difference in core clock.

View attachment 224756

3080 is undervolted @ 1815mhz -- not bad for a "6" core. Biggest jump in Timespy score was also ram+cache driven. Basically miniscule difference from 5.0 to 5.3 core (except the massive echub). Huge double digit % gains on cache and ram to 4000 Gear1. Platform seems starved of memory/cache with heaps of clock/ IPC in the tank.

I've been wondering about that. DDR5 is in its infancy, ADL so far seems to be able to keep increasing IPC if you can keep it fed.

There are some folks on overclock.net that have it up to DDR5-6400 C30. This memory because of the speed actually has lower latency than my DDR4-3200 C16 :

 

[lexluthermiester]

Reports of AL having issues with some denuvo games, is this true?

It is true. Denuvo will not run on Alder Lake, regardless of OS version.

I have my 12900k limited to 200w package power, it doesnt seem to drop the score much.
Actually, i feel like the sweet spot is around 180w package power.

180 Package Power maxes around 70c and doesnt drop scores much more than this.

I havnt even attempted to OC it because its so damn fast anyway i see no need.

View attachment 224723

These are the experiences most people are having that I have seen. The "sweet spot" seems to vary a bit from sample to sample, which seems like a silicon lottery scenario. Alder Lake performance is outstanding, no doubt, but Intel seems to have pushed the architecture close to it's limits. So OCing is showing limited returns.

 

[Outback Bronze]

Z690 Prime D4 with 12700KF in my HTPC : )

Had to make sure the old Noctua was going to be compatible. It wasn't a snug fit but works.

Made sure paste was squished.

Heatsink on : )

My HTPC Case.

All Installed : )

Noticed temps are quite hot but the Noctua is a HTPC one which I cant remember the name of so its all good.

I'm running at 4.4Ghz All Core with HT off and no E cores atm. Temps under RealBench stress are about 60c gaming 50c. This is with a -0.1600v undervolt everything stable with gear 1, 3467Mhz 15,15,15,28,2T atm

Now I think the bios needs some work. As soon as I take it off "Sync all Cores" it wont boot windows. Even if un-sync them and they are all @ 4.4Ghz. I would like 2 cores running @ 4.8Ghz with 6x @ 4.4Ghz

Running Window 10 atm with no real issues except for the OC bios problem stated above.

Cheers.

 

[lexluthermiester]

Made sure paste was squished.

That paste needs to be spread across the entire IHS. This is important. Alder Lake dies are long.

 

[Outback Bronze]

That paste needs to be spread across the entire IHS. This is important. Alder Lake dies are long.

That was just to test to make sure it was going to make contact which it did. That HSF is not made for Socket 1700.

The next application I did properly.

 

[lexluthermiester]

That was just to test to make sure it was going to make contact which it did. That HSF is not made for Socket 1700.

The next application I did properly.

Ah ok.

 

[PolRoger]

That was just to test to make sure it was going to make contact which it did. That HSF is not made for Socket 1700.

The next application I did properly.

That looks like a NH-U9B variant?? You need to get the correct new Noctua NM-i17xx bottom bracket adaptor for you cooler. The original 115X bracket might work for short term temporary use but there are some changes to ADL socket specs vs 115X. There are two types of new Noctua brackets available depending on actual cooler revision etc. I got two of them go with my two older Noctua coolers.

Check the Noctua website for the correct match to your specific cooler version.

NM-i17xx-MP78 mounting-kit (noctua.at)

NM-i17xx-MP83 mounting-kit (noctua.at)

 

[Outback Bronze]

You need to get the correct new Noctua NM-i17xx bottom bracket adaptor for you cooler.

I doubt it. Been like this for a few days with temps staying the same. Its been screwed on pretty tight so cant see how its going to move? Ill keep monitoring it though. Cheers.

 

[GerKNG]

made two videos of the down clocking issue including the bios settings... (i tested per core Overclocks, specific ratios per core, everything on auto/enabled/disabled. manual voltage/current limits in the IA VR Configs)

first one is 100 Mhz above the SC Boost (5 Ghz) and it down clocks to 4.8 to be below the 4.9 max turbo.
second one is 4.8 Ghz (to be 100 Mhz below the SC boost) and it imediately stops down clocking.

12600k frequency Problem #1 (5Ghz) - YouTube
12600k frequency Problem #2 (4.8Ghz) - YouTube

 

[RandallFlagg]

I doubt it. Been like this for a few days with temps staying the same. Its been screwed on pretty tight so cant see how its going to move? Ill keep monitoring it though. Cheers.

Would be interested in seeing what HWInfo shows for power and temp while running CPU-Z Bench. If I move to Alder Lake it would likely be the mATX version of your motherboard.

 

[phanbuey]

Grabbed top spot for 12600K CPU in timespy for now... People still waiting on their deliveries

 

[AlwaysHope]

...
Made sure paste was squished.
...

Oh my... I would never do that with TIM . Could never be certain no tiny air pockets got stuck when screwing down the HS.

 

[Outback Bronze]

Oh my... I would never do that with TIM . Could never be certain no tiny air pockets got stuck when screwing down the HS.

Not sure I understand you there matey? Are you talking about the type of application I applied? I do a line down the middle of the CPU.

I was just testing to make sure the TIM "did" get compressed between the CPU and HSF as the bracket for the Noctua is not made for socket 1700 and so far its working sweet once I did a proper application of TIM.

 

[lexluthermiester]

Oh my... I would never do that with TIM . Could never be certain no tiny air pockets got stuck when screwing down the HS.

Air pockets are almost a myth these days. It's just not going to happen.

 

[RandallFlagg]

View attachment 224875

Grabbed top spot for 12600K CPU in timespy for now... People still waiting on their deliveries

That's an excellent score given that the #1 non HEDT (two memory channel) score on the planet is only getting a 25% higher score :

SkatterBencher got 5.7Ghz on a 12900K with open loop water :

 

[phanbuey]

Once I get some more time I will try to find a golden core or 2 to pump to 5.5 ghz...

 

[RandallFlagg]

ram 4x8gb 4000MHz Gear 1 (2000% kuhru ram test stable) - 17 18 18 37 | TRRDS/TRRDL/TFAW = 4/5/20 | TRFC= 400(200ns) - no RTL/IOL tuning yet.

Hey what brand of memory is that?

I think I'm going to upgrade my storage to a single m.2 2TB and my RAM to DDR4-4000, was looking at some Trident Z F4-4000C18D-32GTZR. Those timings you have are pretty tight compared to the memory I was looking at though.