Does a 12700K and Alder Lake CPUs run much hotter with DDR5 RAM than DDR4

[Wolverine2349]

I just switched out my 12700K from a DDR4 Asus Tuf DDR4 board and put it in an Asus Prime Z690-A DDR5 board and 6000MHz DDR5 RAM

Same exact cooler and case and etc..

On the Asus Tuf Z690 WiFi D4, running OCCT Variable Large Dataset Test, temps peaked in high 70s and averaged in mid to high 60s

Now with an Asus Prime Z690-A DDR5 board, running OCCT Variable Large Data Set temps already peaked at 88C and average very high 70s and spike in low 80s.



Both boards have same layout and almost VRMs with the Prime A being slightly better VRM I think and BIOS is the same.

Same LLC 7 and same VCORE 1.275 and same BIOS settings as they are both ASUS boards. E cores and HT disabled on both. Both had 5GHz static all core OC. Same exact i7 12700K CPU.

Does DDR5 stress IMC of these CPUs for 10C or more higher temps?

I am surprised it would given Gear 2 Memory controller with DDR5 is only running at 1500MHz (3000MHz RAM non DDR rate) vs 1800MHz with the DDR4 Gear 1 mode.

Any info on this.

I also had a similar experience on another board.

Is it a thing that DDR5 at higher speeds makes CPU much hotter? Would lower DDR5 speeds help?

 

[Dan_D]

I doubt it. However, I don't think its been properly tested.

 

[DooKey]

I would argue your HSF probably needs to be reseated or the DDR5 board may be overvolting your CPU compared to the DDR4 board.

 

[Dan_D]

I would argue your HSF probably needs to be reseated or the DDR5 board may be overvolting your CPU compared to the DDR4 board.

Some boards also have things like Multi-Core enhancement (ASUS) which can be effecting your temperatures as well. But I'd agree that its likely an issue with the heat sink installation or application of thermal paste.

 

[Wolverine2349]

I did downclock the DDR5 RAM to 4800MHz and temps dropped 8-10C. Not quite as cool as DDR4, but better and close. Is it possible the sticks heat spreaders run hotter near large air cooler causing CPU temps to spike??

Maybe the Crucial Ballastix DDR4 has cooler running heat spreaders??

Think I am going to go back to DDR4 in that case as non fast high timings DDR5 is not worth it. Problem is the options of DDR4 boards are limited.

 

[Wolverine2349]

I would argue your HSF probably needs to be reseated or the DDR5 board may be overvolting your CPU compared to the DDR4 board.

I had tested 2 boards this and a Rog Maximus Hero with T.Force and Trident X RAM 6000MHz and both ran at least 10C hotter under load in OCCT Variable Normal Large Data Set test with scary peak temps.

 

[JSHamlet234]

I did downclock the DDR5 RAM to 4800MHz and temps dropped 8-10C. Not quite as cool as DDR4, but better and close. Is it possible the sticks heat spreaders run hotter near large air cooler causing CPU temps to spike??

Maybe the Crucial Ballastix DDR4 has cooler running heat spreaders??

Think I am going to go back to DDR4 in that case as non fast high timings DDR5 is not worth it. Problem is the options of DDR4 boards are limited.

No. There wouldn't be enough of a difference in heat from the RAM to make such a difference. 1 degree maybe, at most.

What's happening is that the faster RAM is allowing your CPU to crunch OCCT Large without stalling (as much).

Try a test that is less memory intensive than OCCT large and see if there is a difference in temperature with the faster RAM. My guess is that there won't be.

 

[robbiekhan]

Some boards also have things like Multi-Core enhancement (ASUS) which can be effecting your temperatures as well. But I'd agree that its likely an issue with the heat sink installation or application of thermal paste.

Gigabyte has this option set to auto by default but auto is off it seems until the BIOS deems it worthy? If I manually set it to enabled then Cinebench R23 takes my 12700K to 100 degrees fairly quick. With it on Auto the temp does not go above the mid 70s. The score difference in multi core is only 1000 points. No difference in single core which is expected. It's not a feature I care to leave on for such little positive gain!

 

[Dan_D]

Gigabyte has this option set to auto by default but auto is off it seems until the BIOS deems it worthy? If I manually set it to enabled then Cinebench R23 takes my 12700K to 100 degrees fairly quick. With it on Auto the temp does not go above the mid 70s. The score difference in multi core is only 1000 points. No difference in single core which is expected. It's not a feature I care to leave on for such little positive gain!

It's auto-overclocking of a sort. Essentially, the motherboard manufacturer ignores certain thermal and power limits the CPU manufacturer has defined and uses their own. The gains aren't huge but they are usually consistent. The problem is that these settings are pretty conservative. They have to be or they wouldn't run on the majority of CPU's. That's why you can see some gains, but nothing earth shattering. The silicon of these CPU's is pretty close to the edge of what it can do. There isn't a lot of room for improvement.