Prolonged use of DDR3 at stock voltage on Skylake can damage CPU says Intel

[colinmcr]

Thought I'd share this

Quote

"Intel Corp.’s latest “Skylake” processors officially support only DDR3L and DDR4 memory, but there are motherboards for the new chips that can also use DDR3. While the chips can work with previous-gen memory, prolonged usage of such dynamic random access memory (DRAM) can damage microprocessors, according to Intel."

Full story here

http://www.kitguru.net/components/m...emory-at-default-voltages-can-damage-skylake/

Source story from Tomshardware

http://www.tomshardware.com/news/skylake-memory-support,30185.html#xtor=RSS-181

 

[TubbyGreg]

Huh, that's interesting. I'd probably just go with DDR4 just to be on the safe side

 

[horrorwood]

** Edited ** There is no direct quote from Intel that says 1.5v RAM will damage it.

There is 1.4v ram on their QVL list right here: http://www.intel.com/content/www/us/en/gaming/xmp-for-core-processors.html

 

[mmj_uk]

That's just a list of XMP profiles (eXtreme Memory Profiles), eXtreme being the key word.

If the Skylake memory controller is designed to run at 1.2-1.35V and you stick 1.5V XMP memory in it then you're running it out of spec... whether that will cause degredation or damage I guess only Intel's engineers really know. We had the same sorts of discussions when SB launched with 1.65V memory, I guess everyone just made the move to 1.5V memory anyway.

 

[Silent_Scone]

Anything over 1.5v long term I would say has the potential to cause long term degradation - maybe. There is literally no way of knowing for sure.

 

[colinmcr]

I wonder when the reporter or tester who did the original story, did actually contact Intel? In his article he says, he spoke with Intel. We have no actual proof of that and surely the correct thing to do as a reporter is to quote from a named source.
The story then becomes like a Chinese whisper and becomes "offical" even though it may not actually be.

I searched all Intel press releases and news and could not find anything relating to this directly from Intel either.

So in this case, us end users read an article and take it as given (mostly). I have no opinion as to the validity of the claim, all I can do is share with my fellow enthusiast's in the hope that none of them suffer the fate of this "possible" flaw.
I'd certainly prefer to err on the side of caution than to ignore the possibility.

 

[dreamcat4]

it is because the memory controller is on the CPU, and the memory controller must run at same voltage as the ram...

So as the memory controller is on 14nm silicon it's effectively the same situation as if you were to try overclocking your Skylake CPU above 1.4 volts. i.e. not recommended.

That said the IMC runs at it's own voltage. So the rest of the chip would remain OK. But if damaged the entire CPU will be useless anyway. As it would have no longer have anyway to read / write to main memory.

On a positive note, if Intel think 1.35v is an OK voltage to run the IMC at (long-term). Then maybe it's also a safe-ish voltage to be used as a general CPU overclock. Given the same 14nm transistors in both IMC and throughout the rest of the CPU.

Although perhaps the lifetime of the product would be lessened by some? years. How many is rather hard to say without long-term testing.