Am I silly for insisting on 1.2v RAM?

Zangmonkey

Supreme [H]ardness
Joined
Jul 6, 2005
Messages
5,065
DDR4 standard is 1.2v but tons of enthusiast sticks list 1.35v

When I plan to just use XMP I use the chart to choose the best speed/latency combo I can find but *only at 1.2v*

Am I being silly here?
 
Last edited:
It does help as stock voltage does keep my sticks from being stable at XMP. But I bought a 6600k so voltage adjustment was always in the cards. Come to think about it, having to add some voltage to get stable memory was an early push into overclocking for me. I had a set of sticks I had to remove all but 1 just to bump up the voltage up in the bios.
But yes I can see not wanting any kind of adjustment from stock as very important. I still keep an old LAN card around because I know it works without drivers.
 
yes that is silly. the 1.2v is for the slowest speeds only and anything decent is going to require 1.35v. why fret over .15v?
 
yes that is silly. the 1.2v is for the slowest speeds only and anything decent is going to require 1.35v. why fret over .15v?


I can get RAM with a higher XMP speed setting at stock voltage.
.15v is a 12.5% increase in voltage.
 
then buy it and then what is the point of this thread?
and...
worried about $0.05 a yr?! unnoticible temp difference? I don't get it and I think is silly, like you asked.
 
If the memory comes rated at 1.35v, I wouldn't worry about it. Factory overclock, like Sven said.

You wouldn't worry about buying a GPU that runs a higher clockspeed than the reference cards of the same GPU; for all you know, they fiddled with the voltage to get there. You don't worry because the warranty and design of the card covers operating in that state.

1.35v memory is the same.
 
The concern would not be the longevity of the memory as that is what it is in spec for. Although I am skeptical on how much testing and engineering and built in tolerances are actually built in "OCed" memory kits as many of these cannot even hit their advertised specs reliability at the advertised 1.35 voltage.

What the issue is the impact on the the IMC in the CPU as that officially is only rated for 1.2v +/- 5% for DDR4. And yes if you look in the databook for Skylake the table entries for this spec is separate for DDR3 and DDR4. DDR3 is 1.35v +/- 5%.

I remember back when Sandybridge was released and official DDR3 voltage support was dropped down to 1.5v the official comment from Intel and some other partners was that 1.65v DDR3 can result in damage to the IMC. Intel's response when queried for Skylake and DDR3 was also that anything higher than 1.35v was also out of spec (http://www.tomshardware.com/news/skylake-memory-support,30185.html). As far as I know however no one has pressed them on what the situation is with 1.35v DDR4.

I think there is some assumption involved that internally Skylake can handle 1.35v DDR4 since it officially supports 1.35v DDR3. Also that Skylake supports DDR4 up to 4133mhz.

I can get RAM with a higher XMP speed setting at stock voltage.
.15v is a 12.5% increase in voltage.

Limited to DDR4 2800 CL16 I believe, at least for 2x8gb (which is really the smallest you'd want to go with). If you want anything faster they are only available officially at 1.35v.
 
Last edited:
I think there is some assumption involved that internally Skylake can handle 1.35v DDR4 since it officially supports 1.35v DDR3. Also that Skylake supports DDR4 up to 4133mhz.
Yep this.
The connections are the exact same and will be subject to the same electrical tolerances.
1.4175V is within 5% tolerance so a fair bit of overclocking headroom exists.


Speed support is a different fish, trying faster ram speeds cant cause harm, only the voltage.
There is a dual limit when overclocking both CPU and GPU to the max though.
Either being maxed can cause the other to become unstable.

When my 6700K is at 4.6GHz I can run my 3GHz ram (as sold: 15, 17, 17, 35, T2) at 3.6GHz with 1.4V (17, 18, 41, tRFC = 380, T2).
Boosting my CPU to 4.7GHz I cannot get 3.6GHz ram stable.
Instead I drop to 3333MHz and much tighter timings at 1.39V (16, 16, 16, 31, tRFC = 320, T1).
It will do 3333MHz at 1.35V with looser timings, but why do that :)
 
Same boat as you my dude. I'm making a Plex/Media/File server/system and I'm adamant on 1.2V even though it makes no difference in the end when the electricity bill comes.
 
Back
Top