Maximum safe VTT on SB-E?

[AdamK47]

Yes, I'm going to actually ask an overclocking question in this forum for a change.

Purchased an ASRock Extreme11 and 32GB of G.Skill Ripjaws Z 2400 DDR3 memory recently. I've been able to get 4.5GHz stable on my 3960X with the board (up from 4.3 on my previous Asus P9X79 Deluxe). Getting 2400 out the memory is the troubling thing at the moment. The XMP 1.3 profile seems fine except for the VTT. The XMP profile calls for 1.2V on the VTT. This is totally unstable and will fail LinX AVX within minutes. If I increase it to 1.3V it will run LinX AVX for about an hour before failing. Is 1.35V or 1.4V safe VTT for SB-E? Anyone running a similar setup? It will run fine at 2133 at 9-11-11-28 timings with only 1.05V VTT. Part of me wants to leave it there, but the other part of me wants the full 2400 DDR3 speeds.

 

[BonzaiDuck]

Yes, I'm going to actually ask an overclocking question in this forum for a change.

Purchased an ASRock Extreme11 and 32GB of G.Skill Ripjaws Z 2400 DDR3 memory recently. I've been able to get 4.5GHz stable on my 3960X with the board (up from 4.3 on my previous Asus P9X79 Deluxe). Getting 2400 out the memory is the troubling thing at the moment. The XMP 1.3 profile seems fine except for the VTT. The XMP profile calls for 1.2V on the VTT. This is totally unstable and will fail LinX AVX within minutes. If I increase it to 1.3V it will run LinX AVX for about an hour before failing. Is 1.35V or 1.4V safe VTT for SB-E? Anyone running a similar setup? It will run fine at 2133 at 9-11-11-28 timings with only 1.05V VTT. Part of me wants to leave it there, but the other part of me wants the full 2400 DDR3 speeds.

By VTT I assume you mean the same thing as what is called VCCIO. Your quoted voltage range of consideration seems to match the spec for VCCIO.

For the regular Sandy Bridge rig -- the "K" chips -- The stock value could be just below 1.00V. Several guides I read were pretty clear that you could boost the voltage to 1.2V without much worry, and that it was possible to run it up as high as 1.35V (but you would then be "taking your chances.")

In order to overclock som DDR3-1600 G.SKILL (4x4GB) to run at 1866, I had to bump up my VCCIO to about 1.12V as I recall.

My advice here -- and I'm a cautious type of person -- try loosening the timings at the desired 2400 speed, and see if you can keep the VCCIO (or VTT) below 1.25V. The lower, the better.

Otherwise, what's wrong with running them at the 2133 speed and timings? Also -- how far did you boost the RAM voltage? How much safe room do you have left for increasing the RAM voltage?

Maybe the spec for the SB-E is more tolerant, but I don't see why it should be.

 

[HutchinsonJC]

I see questions more specifically about vtt, but what voltage are you running the RAM itself at? I'm pretty sure the more sticks of ram you try to use the more likely you'll need to bump the voltage up to keep stability. Usually it's in the ballpark of 0.01 to 0.02 volts. My ram is spec'd 2133 @1.5 but I run it at 1.52v.

 

[AdamK47]

The memory is at 1.65V.

 

[Grooveriding]

I wouldn't go over 1.25, preferably 1.2.

http://rog.asus.com/forum/showthread.php?19926-3930K-C2-Degradation

Anecdotal, but doesn't seem motivated out of trying to spread FUD. Probably better to loosen memory timings than run such a high VTT.

 

[AdamK47]

Tried 1.355V on VTT overnight. Failed LinX AVX with 28GB memory allocation. I won't be trying any higher. Moved it back down to 1.05V with the memory back at 2133. I guess this memory, motherboard, CPU combination just will not do 2400.

 

[HutchinsonJC]

The memory is at 1.65V.

Have you tried bumping the voltage on the ram slightly? 1.66? 1.67?

I've seen it time and again mentioned that more sticks of ram can often require a small bump in voltage to maintain stability than fewer sticks or ram would require.

Anyway, one of the links I use as a reference has this for being possibly relevant to you /shrug

http://rog.asus.com/46212011/rampag...extreme-uefi-bios-guide-for-overclocking-2/3/

VTT CPU Voltage & 2 VTTCPU Voltage: To let users over-volt part of the CPU VTT that affects OC without over-volting the part of VTT that does not affect OC, we split up VTT rails to VTT and Secondary VTT.


VTT over-volting helps with DRAM and BCLK Overclocking. Secondary VTT does not. However, try to maintain at most a 300mv delta between both voltages for stability. 1.40v is fine on these CPUs. 1.35v is sufficient most of the time.

VTTDDR Voltage: Leave at Auto unless experiencing instability with memory. As a starting point, set to 50% of DRAM voltage. So if DRAM voltage is 1.65V, then set ~ 0.825V here.

 

[AdamK47]

I've been able to get 2400 stable now. Had to update the UEFI on the Extreme11 to 2.0. Previous attempts using version 2.0 failed due to video signal issues. Getting two GTX 690s fixed that.

For 2400, I used the XMP profile for all timings and set VCCA and VTT voltages to auto. The XMP profile then set VCCA to 1.200V and VTT to 1.118V.

Used LinX AVX with 11 threads and 28GB of memory allocation. Passed 11 hours of stress testing.