Sign in with
Sign up | Sign in
Your question
Solved

Is Undervolting Safe for Memory?

Tags:
Last response: in Memory
Share
August 7, 2011 3:03:12 AM

Hi all,

I have a G.Skill 8GB Kit (F3-12800CL7D-8GBXH).

The RAM specs are 7-8-7-24-2T rated @1.6v. Well, I wasn't aware that SB should be using 1.5v when I bought the kit. So I'm a little worried (mostly about $$ thrown away that could have been used for something else). My own fault for not really looking into the matter. I understand that running higher DIMM voltage also shortens the lifespan of SB CPU. But not truly known. Am I correct or wrong in what I've searched, read and understood?

I actually undervolted it using the same RAM specs. Basically DIMM voltage set @1.5v (initially @1.55v) with 7-8-7-24-2T. I ran memtest for ~8 hours and going to 8 passes without any errors. System POST, boot and loads fine. I haven't played any games yet or played any multimedia yet. So I'm not sure if it'll crash there.

I have played games and whatnot on it using 1.55v and everything seems fine. I haven't observed any odd behaviour going on. System isn't a 24/7 machine, though I may run it 2-3 straight days before turning it off (mainly when downloading games from Steam, etc).

I guess I started getting concerned on long-term effects this may have on any of my components. Should I try to change the timings to what JDEC has for 1.5v instead (1600 @ 9-9-9-28)? Keep at where it is now? Should I really be concerned?

TIA.
August 7, 2011 1:26:55 PM

No, setting it to lower voltage is not posing a dander to the RAM. It however may be impossible to reach the specified timings. And under some value it will stop working at all :) 
m
0
l
a b } Memory
August 7, 2011 1:50:09 PM

werner123 - "The memory voltage actually has nothing to do with your cpu, "
Not correct. The amplitude of the memory signal (Pulses) is applied to the CPU.
Intel Data sheet specifies a max voltage of 5.575 for the Ram. ??SB has a memory controller that is tighter than prevous cpu (Ix-XXX)

That said, Not sure that that is backed up with hard fact. I also have the 1.6 Volt DDR3 1600 CL7 and have left it at the default value. That is only 0.025 V over stated max - 6.7% vs spec of 5%.

As to operating the ram at a lower voltage, if it is stable, go for it. I would probably not go below the 1.50 point. And yes there is also a min value that , For SB it may be 1.425
m
0
l
Related resources

Best solution

a c 347 } Memory
August 7, 2011 2:28:12 PM

Putting blinders on.

The Sandy Bridge can run 1.65v RAM and there's no proven cases of 1.65v harming the Sandy Bridge at all. YES, I've had this argument even with our Intel guy. ALL of the specs DIMM 1.5v, IMC voltage, you name it are clones of the LGA 1156, LGA 1366 and now LGA 1155 - no differences digging DEEP into the Intel Specs. I get the ooooh Sandy Bridge HAS to have 1.50v DIMM -- well then so did the others before it. When the Sandy Bridge was developed ONLY 1.65v high frequency RAM existed. The Spec for DIMM Voltage per Intel is 1.57v; so IF +0.03v is the Doomsday then a vCore of 1.35v~1.40+ is the LAVA of Hades -- then the 'K' CPU is a myth because you couldn't OC it at ALL adhering to Intel Spec. ** This is where my discussions of Voltages breaks-down even with our Intel guy. **

Next, you only are mentioning the DRAM Voltage and the CAS Timings -- what about the DRAM Frequency? I assume that it is set properly DRAM Frequency -> 1600 MHz or Memory Multiplier -> 16. Also, ooooh the Sandy Bridge Spec is DDR3-1333 MHz so is your CPU going to ExPlOdE??!! Hell no!!!

IF your RAM is running Rated DDR3 1600 MHz 7-8-7-24-2N @ 1.50v with NO Memtest errors then terrific leave it be, otherwise run full Rated DDR3 1600 MHz 7-8-7-24-2N @ 1.60v; both the higher Frequency and lower CAS generally require additional power to BOTH the DIMM and the CPU IMC {QPI/VTT/VCCIO} 1.15~1.20v.

OC your CPU is 10X+ maybe 50X harder on the CPU than 1.50v vs 1.60v DIMM.

Share
August 7, 2011 2:58:23 PM

Thanks all for the replies.

I am running RAM @ specs but with lower voltage.

1600 @ 7-8-7-24-2T @ 1.5v. Memtest showed no errors running it for 8 hours (7 completed passes and partially into the 8th pass). So I should be good to go. :) 
m
0
l
a c 347 } Memory
August 7, 2011 3:21:05 PM

neograndizer said:
Thanks all for the replies.

I am running RAM @ specs but with lower voltage.

1600 @ 7-8-7-24-2T @ 1.5v. Memtest showed no errors running it for 8 hours (7 completed passes and partially into the 8th pass). So I should be good to go. :) 

No Errors = No Changes needed.

Problems later ie Gaming the first place I'd look would be the RAM.
m
0
l
August 7, 2011 3:52:19 PM

jaquith said:
No Errors = No Changes needed.

Problems later ie Gaming the first place I'd look would be the RAM.


jaquith is right, although keep one thing in mind (which I learned from my experience and from another person replied to my other thread for RAM issues):

-- If you ever face some application/process crash or BSOD then try increasing RAM voltage as in my case my RAM was exactly the same timings and speed as yours but I have 2 (2x2GB) kits thus my all 4-RAM slots on motherboard are used and I had to increase RAM voltage 0.03v from its rated 1.6v to stop some random application/process crash during normal system use in OS/Windows.
m
0
l
a c 347 } Memory
August 7, 2011 4:23:01 PM

I originally deleted 'my feelings' - in short 'I' don't want/need something hanging over my head to 'hit it' later down the line. With the OP's set IMO DRAM Voltage -> 1.60v AND QPI/VTT/VCCIO -> 1.20v. All within safe margins and 'Peace of Mind' ;) 
m
0
l
August 7, 2011 5:59:42 PM

I've been running this RAM initially at 1.55v with no problems (gaming and otherwise). But, yeah, it has occurred to me that should anything go 'wrong' it will most likely be the DIMM voltage running below what is spec'd.

jaquith, thanks for the info you have provided.

I'll see how this goes, and maybe just up it to 1.6v and forget about the entire thing.
m
0
l
August 7, 2011 6:00:01 PM

Best answer selected by neograndizer.
m
0
l
!