Corsair Vengeance LPX has two XMP profiles, which to use?

alpha1172 · Feb 23, 2018

Hi guys! I use this RAM kit: https://goo.gl/V9aLJN

I've been running it at the default XMP profile since I got it, which is 2133 MHz 13-15-15-28 1.2V. I've recently discovered that there's another XMP profile, which is at 2400 MHz 14-15-15-32 1.35V.

I'm guessing the second one is faster, but I noticed that it is now running at 1.35V instead of the JEDEC standard 1.2V. I'm always looking for the maximum performance of this new computer that I just built, but it's also supposed to last quite a few years (at least 5).

Should I stick to 1.2V for longterm reliability or is 1.35V not a problem?

Thanks!

Dunlop0078 · Feb 23, 2018

What CPU are you using? Pretty much any CPU that supports DDR4 should be fine with 1.35v, I have seen intel DDR4 memory controllers run a lot higher than that, some really high speed kits use 1.4v or higher in the XMP profile. I personally use 1.35v DDR4 voltage in my kaby lake system without issue.

That kit runs totally stable at 2400mhz 1.35v? If so then go ahead and try manually lowering DDR4 voltage, maybe 1.3v play some games BF1 is particularly good at identifying memory instability from my experience. A program like super pi 24 million is also good for testing ram stability. If it is stable at 1.3v try even lower, it may even run 2400mhz at 1.2v stable.

alpha1172 · Feb 23, 2018

Dunlop0078 :

Oops forgot to include my system:

Gigabyte Z370 HD3
Intel i5 8400
Zotac GTX 1070

I ran 8 passes of Memtest v7.5 on the 2133 CL13 XMP Profile when I first got the ram. I switched to the second XMP profile tonight, but I haven't run memtest on it. It boots though, I'm using it right now. I looked on Corsair's website and they say the second profile is, and I quote

"Why are there two XMP profiles on my Corsair DDR4?
We include a pair of XMP profiles instead of just one for users who want to control how much power is consumed by the memory. The first XMP profile runs the DDR4 at its specification of 1.2V, while the second offers a higher speed at the cost of bumping the voltage to 1.35V. The first profile, then, is officially supported, while the second is not and instead offers a baseline of what the memory should be able to achieve. "

I'm assuming /should/ be able to achieve means they haven't validated this kit to work at those speeds (because if they did I'm guessing they'd sell it as a 2400 CL14 kit?)

Dunlop0078 · Feb 23, 2018

Memtest is not so great for testing stability of overclocked RAM from what I have read. I use memtest to check for defects in a new kit, or verify a hardware problem with an old kit or a CPU memory controller.

Yeah pretty much, but just because it was not validated for 2400mhz does not mean it wont do it. I would imagine pretty much any kit rated for 2133mhz 1.2v will hit 2400mhz stable with the same or looser timings at 1.35v.

You can change the DDR4 voltage manually with the XMP profile enabled if you want, that is what I would personally do, I bet that kit will run stable at 2400mhz with less than 1.35v.

alpha1172 · Feb 23, 2018

Dunlop0078 :

I will try to play some BF1 to test the stability as you mentioned, but I will also try memtest.

If it's stable, should I:
A. Tighten timings
B. Decrease Voltage
C. Try to go for 2666 and beyond?

Dunlop0078 · Feb 23, 2018

You can try 2666mhz, might work. I doubt you will be able to get higher than that with that kit unless you use more than 1.35v which I would not recommend for longevity of the CPU memory controller. 2666mhz will likely require 1.35v, if you stay at 2400mhz I would try lowering the voltage and keep the timings as is.

If 2666mhz is not stable at 1.35v 14-15-15-32, try 15-15-15-32, if it's still not stable try 16-16-16-36. Increasing speed will almost always result in more performance in the vast majority of applications, assuming you don't have to slack up the timings a hell of a lot to get that speed stable.

alpha1172 · Feb 23, 2018

Dunlop0078 :

I'll try to do that tonight!

Thanks! 1.35V is okay for both the ram and the memory controller?

Dunlop0078 · Feb 23, 2018

alpha1172 :

It certainly seems to be. I cant say for sure until there are skylake, kaby lake, and coffee lake processors that have been running that voltage for 5 years. If a bunch start failing then we will know for sure. I would say it's highly unlikely, intel DDR4 memory controllers for the most part seem pretty good so far. The RAM sticks themselves will take more than 1.35v without issue, I wouldn't worry about the sticks at all at that voltage.

Forgot to say. If you try 2666mhz 14-15-15-32, the system may not POST meaning you will just get a black screen and no boot. If it does wait a minute, the system should try to restart then a screen should come up saying failed to boot, overclock failed, failed to post, something along those lines and give you a prompt to enter the BIOS. If it does not restart, just hold the power button and hard reset it, this wont corrupt anything as nothing should be loaded into RAM at the time. Then enter the BIOS and increase the timings and try again.

alpha1172 · Feb 24, 2018

Dunlop0078 :

Okay I don't really know what I'm doing. I read about overclocking the i5 8400 by increasing the base clock. It's not much, and the maximum intel allows (for locked CPUs) is 102.7 MHz. I set it at that, and now I get 3.904 GHz on all cores under load. Pretty decent 100 MHz overclock.

Anyway, that also increased my RAM speeds by a bit. I was playing around with 100/133 base options and multipliers but my RAM kept going from 2600 to 2666 to 2737 to 2600 randomly after every boot.

What I ended up doing, was I looked at all the timings of the 2133 CL13 XMP profile. I then set the RAM speed multiplier to 24, so now the speed is 2464 MHz. I then applied the exact same timings of the 2133 profile, and viola it works. Obviously the 2464 CL13 would be faster than the other 2400 CL14. This is still at 1.35V.

With both BCLK overclock and this nice little memory tweak, my Cinebench R15 score went from 895 to 955. It's always nice squeezing out the most performance, especially from a budget machine.

alpha1172 · Feb 24, 2018

Btw when I say copy the timings I mean I copied everything including the subtimings.

I tried a 2666 14-15-15-32 one and left subtimings on auto but it ended up being slower than the 2400 CL 14-15-15-32 XMP profile (cinebench).

When I copied the exact timings of the 2400 CL14 XMP to the 2666 one, it wouldn't post. Is there really a point in increasing speeds if the timing increases anyway? Other forums say to calculate your latency in nanoseconds. And the best I've gotten so far is the 2400 CL13 one.

I'm also not sure if any of this even matters. I put back the original 2133 CL13 XMP profile, though now it's 2190 because of the BCLK overclock. I tried it in cinebench and I got 948.

So the difference between 2133 CL13 and 2464 CL13 is 7 points? Haha maybe I'm using the wrong program to test. Also for stability, all I know as of now is whether or not it posts lol.

Update: I got it to 2600 MHz (though Task Manager says 2584). It's still using the exact same 2133 CL13 timings (13-15-15-28 etc etc). How exactly do I stability test this?

Update 2: My 2600 CL13 timings (mentioned prior) survived 10 passes of Intel Burn Test v2.5 @ 14.5 Gb Memory usage.

Search

Corsair Vengeance LPX has two XMP profiles, which to use?

alpha1172

Distinguished

Dunlop0078

Titan

alpha1172

Distinguished

Dunlop0078

Titan

alpha1172

Distinguished

Dunlop0078

Titan

alpha1172

Distinguished

Dunlop0078

Titan

alpha1172

Distinguished

alpha1172

Distinguished

TRENDING THREADS

Latest posts

Moderators online

Share this page