Discussion Alder Lake - Builders Thread

[AdamK47]

This thread is for those of us that own or are looking to get an Alder Lake CPU.

Bought mine on release day. 12900K. Memory selection was limited to only Crucial DDR5 4800 at the Micro Center I went to. Bought two 2 x 8GB kits. Motherboard is a Gigabyte Z690 Aorus Master. I did buy a 360mm Lian Li Galahad 360mm AIO. Ended up returning it since the Lian Li Socket 1700 backplate that I bought separately worked well with my NZXT Kraken X73.

 

[AdamK47]

You must put personal commentary alongside any link or picture when posting in the CPU forum. -Moderator Shmee

 

[DrMrLordX]

That is very blue. Like late 90s/early aughts blue LED mania blue.

 

[AdamK47]

Blue is the best. Everyone knows this.

I've been running the 12900KS at 5.3GHz all PCore (5.5GHz two PCore) and 4GHz all ECore. Ran Prime95 for 4 to 5 hours. So far it has been stable. Runs at less voltage at 5.3GHz than the 12900K required for 5.1GHz. Runs cooler too.

 

[igor_kavinski]

Runs cooler too.

What are the temps and power usage during Prime95?

GB5, if you don't mind please.

 

[AdamK47]

What are the temps and power usage during Prime95?

GB5, if you don't mind please.

Not sure about power usage. I'd have to go back and check. Prime95 4K FFT with 24GB of memory allocated has the top core sitting just at 90C. That's the worst case torture test I have. Of course normal gaming is much lower with a core topping out somewhere in the upper 60s. The same 4K FFT test would hit 100C and throttle on the 12900K. This even with the 12900KS at 5.3GHz and the 12900K at 5.1GHz.

 

[igor_kavinski]

The same 4K FFT test would hit 100C and throttle on the 12900K. This even with the 12900KS at 5.3GHz and the 12900K at 5.1GHz.

This just goes to show how bad typical reviewers are. No one bothered to test this aspect. These are premium bins then.

 

[AdamK47]

This just goes to show how bad typical reviewers are. No one bothered to test this aspect. These are premium bins then.

I know for a fact there was no way the K would last more than 5 minutes in the 4K FFT 24GB test @ 5.3GHz. The KS does it easily.

 

[VirtualLarry]

That's def. some top-bin silicon happening there.

 

[JoeRambo]

Alder Lake had some interesting behavior with AVX clocks, i think i saw Igor claiming it was limiting it to 5Ghz or so, i wonder if that is still the case with 12900KS, HwInfo64 should show effective clocks nicely.

 

[AdamK47]

Alder Lake had some interesting behavior with AVX clocks, i think i saw Igor claiming it was limiting it to 5Ghz or so, i wonder if that is still the case with 12900KS, HwInfo64 should show effective clocks nicely.

It does deal with AVX loads differently. It can be overridden in the BIOS to tell it to have an offset of 0. I keep it at default. At 5.3GHz it will offset the full PCore to 5.0GHz. I turn AVX off in Prime95 for the all 5.3GHz 4K FFT 24GB test. I just looked at the current running test. Seeing around 300W with this test. This is the same wattage when I run it with AVX and it down clocks the PCores to 5.0GHz.

 

[AdamK47]

Ran AIDA64 earlier for a few hours. More of a realistic high level load. I'm going to be "that guy" and show a camera photo of it instead of a screenshot.

Around 250W. Temps do spike into the low 90s every once and a while.

 

[AdamK47]

Played around with the CPU some more today. Tried enabling "Enhanced multi-core performance mode" on my Gigabyte Z690 Master motherboard. Left all other CPU settings at default including voltage and LLC.

The setting is the cheat setting that a lot of motherboard makers like to have enabled by default. It makes the CPU run at its highest turbo at all times. My motherboard did not have it enabled by default. I decided to try it.

The 12900KS has a new feature called "Enhanced TVB". The combination of the setting I just mentioned and Enhanced TVB introduced interesting results. In the BIOS it will show it's running at 5.5GHz at 1.48V. In Windows I don't ever see it running at that. At idle all cores sit at 5.4GHz with 1.4V. On light workloads it will maintain 5.4GHz. Doesn't matter how many cores are used. If it doesn't push the CPU too much it stays at 5.4GHz. It will sit at 5.3GHz all cores with 1.334V using heavier loads. AIDA64 stress test will do this. At even heavier loads, like Prime95 4K non-AVX, it will cause the CPU to bounce between 5.1GHz and 5.3GHz. And of course, AVX loads cause the CPU to go down to 5GHz all core.

The good thing so far is that everything appears to be stable. Ran several tests today. Going to try out some gaming and see if 5.4GHz all core holds.

 

[coercitiv]

I've always been weary of using MC or even XMP for that matter. You never know what voltages get changed when enabling these options, case in point I had a MSI Z370 board set a ridiculous VCCSA just because I had a DDR4 4000 kit installed.

Since the 12900KS has a factory defined voltage curve up to 5500Mhz, I would rather just manually set an 8-core turbo of 5.4Ghz and then check on Vcore and temps, maybe optimize LLC and AC/DC Loadline lower temps or stability if need be. I'd rather finetune or solve my own mess than search all over the UEFI for obscure "automated" changes.

 

[AdamK47]

In my case a stable 5.4GHz overclock is not possible with voltage and LLC tweaks. With multi-core enhancement I can get at least 5.3GHz while bumping to 5.4GHz as Enhanced TVB allows it. With "enhanced multi-core performance mode" enabled, the voltage and LLC settings are not increased.

 

[AdamK47]

GB5, if you don't mind please.

Got around to running Geekbench.

Gigabyte Technology Co., Ltd. Z690 AORUS MASTER - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z690 AORUS MASTER with an Intel Core i9-12900KS processor.

browser.geekbench.com

 

[DAPUNISHER]

If you built 12th gen with a DDR5 board and ram, congratulations. It is aging like fine blueberry wine. In newer games the memory bandwidth is making a difference. The uplift in performance with 6400 versus a typical DDR4 3600 setup, is similar to what AMD owners get with a 3D part over vanilla.

 

[BTRY B 529th FA BN]

If you built 12th gen with a DDR5 board and ram, congratulations. It is aging like fine blueberry wine. In newer games the memory bandwidth is making a difference. The uplift in performance with 6400 versus a typical DDR4 3600 setup, is similar to what AMD owners get with a 3D part over vanilla.




View attachment 102727

It's a nice uplift! I felt like it was going to a new/next gen chip with gaming when I swapped out my 5800X to the 5800X3D

 

[igor_kavinski]

It's a nice uplift! I felt like it was going to a new/next gen chip with gaming when I swapped out my 5800X to the 5800X3D

One thing that I can't seem to get an answer from anyone about. How does the 5800X3D seem in non-gaming use, from a seat of the pants perspective, compared to the dear departed 5800X?

Have you noticed faster application launch times? Remember any hitchiness doing something with the old CPU that you have never encountered with the X3D?

 

[igor_kavinski]

View attachment 102727

Yeah, I like my 12700K with DDR5-7000. It's nothing too special but I think I could live with it for a few years. My mobo isn't built for anything above 7000 so there's probably slight perf uplift left on the table if I pick up a two RAM slot mobo on sale or from someone here in FS/FT few years from now. Or I could use this chart to convince someone, "Hey, my 12700K with DDR5-7000 could probably beat the crap out of a 14900K with DDR4-3600". If the math clicks in his head, he'll be like, oh yeah I want a 14900K/DDR4 beating machine!

 

[BTRY B 529th FA BN]

One thing that I can't seem to get an answer from anyone about. How does the 5800X3D seem in non-gaming use, from a seat of the pants perspective, compared to the dear departed 5800X?

Have you noticed faster application launch times? Remember any hitchiness doing something with the old CPU that you have never encountered with the X3D?

I can't give a fair comparison with day to day usage, however I am thinking of swapping out my main rig 5800X to a 5800X3D chip that's been sitting around for almost a year, BNIB, unopened.

When tim sweeney decided to shut down the UT4 servers the community setup a server to continue playing. The game does launch slower now that launcher is no longer needed but that might not be hardware related. Other than that games launch the same, nothing different in that respect. I'd say it's an overall net positive experience.

 

[DAPUNISHER]

I will answer the Zen 3 stuff in the builders thread for it. This isn't the place for that discussion. I just thought it was an impressive and similar gaming boost. And I feel like the 20% plus uplift makes the 12900K the real MVP of LGA 1700. It can be found for a great price at times too.

Also the 12600KF was going for like a $125, and with DDR5 boards and ram dropping so much in price, that is one hell of a bang for buck gamer right there. And it should do it on a $35 air cooler no problem.

 

[coercitiv]

When I built my 12th gen system (near launch), the added cost from DDR5 5000+, let alone 6000+ would have been massive.

Hardware Unboxed addressed this issue back during the 12900K release, 2 years ago:

• 32GB DDR4 3600 - $139
• 32GB DDR5 4800 - $309
• mobo added cost would have been $15-20 or more

Even when prices settled down a bit, going for DDR5 6000 would have meant the same $200 price delta, if not more. All of this for 20% performance in games today, on a 4090 @ 1080p. If I buy today a new MSI Z790 Pro board + 32 GB DDR5 6000 from the same local shop I bought 2 years ago, the upgrade would cost me ~$270 when factoring out VAT (so that we can compare with the US prices above).

So the worst case for me would be that I spent $70 for an extra Z690 mobo + 32GB of DDR4 3600.

 

[DAPUNISHER]

When I built my 12th gen system (near launch), the added cost from DDR5 5000+, let alone 6000+ would have been massive.

Hardware Unboxed addressed this issue back during the 12900K release, 2 years ago:

• 32GB DDR4 3600 - $139
• 32GB DDR5 4800 - $309
• mobo added cost would have been $15-20 or more

Even when prices settled down a bit, going for DDR5 6000 would have meant the same $200 price delta, if not more. All of this for 20% performance in games today, on a 4090 @ 1080p. If I buy today a new MSI Z790 Pro board + 32 GB DDR5 6000 from the same local shop I bought 2 years ago, the upgrade would cost me ~$270 when factoring out VAT (so that we can compare with the US prices above).

So the worst case for me would be that I spent $70 for an extra Z690 mobo + 32GB of DDR4 3600.

That's fair. Everyone has to decide value and opportunity cost for their situation.

There are other things to consider. DDR5 can carry over to the next build. With new games expanding the lead over DDR4 with the same CPU, LGA DDR5 boards are going to hold better resale value, providing likely better ROI than DDR4 boards. Not important to some, but for those of us active in the trading community it matters. I can have trouble selling old budget and mid hardware without practically giving it away. Good stuff sells quickly at closer to what I paid for it.

For builders now, it provides another good reason to go DDR5. Not just longevity of the standard, but quite significant performance improvements to be had.

 

[VirtualLarry]

For builders now, it provides another good reason to go DDR5. Not just longevity of the standard, but quite significant performance improvements to be had.

How are those Skylake rigs with DDR3 holding up these days, <LOL>.

 

[DAPUNISHER]

How are those Skylake rigs with DDR3 holding up these days, <LOL>.

Are any of those CPUs still alive? IIRC Intel warned regular DDR3 would degrade the IMC right?