Can adding more RAM destabilize a system?

[spotpuff]

I recently added another 4GB of RAM to my motherboard, so now I'm running 4x2GB of DDR3.

I started getting random BSODs and crashes, but it's inconsistent. After running some tests in Windows at my OC'd settings using Linx, Prime95 and a memory testing program, it seems like there's an issue with some of the RAM (but not all of it); 3 prime threads will keep running but one will crash right away and the RAM checker only errors if it includes a certain portion of the RAM.

Decided to try Memtest86+ and on reboot it started errorring immediately. OK, that looks bad, so I reset the defaults in the BIOS and ran memtest again, and it started throwing errors again. I was playing around in memtest's settings to try the "probe" setting under how much memory to test, and the screen output became garbled and the computer seemed to hardlock.

Upon reboot (still at stock settings) memtest ran for 3 hours without any incidents. I'm still running it now (while I'm at work) so by the time I get home it will have been running for 14 hours or so.

Previous to this I had been running at 3.5GHz at stock vcore/vtt/everything (RAM @ 1.66v) but after a few months it appeared that speed was unstable. I updated to a new BIOS but I could no longer use the 21x multiplier, but I settled on 20x170 to hit 3.4 on stock and everything seemed fine (this is with the new RAM); linx for 10 minutes ran fine, etc.

Is it possible that processors require more voltage over time to hit the same clock speeds? Or that adding RAM can destabilize an otherwise ok system? Is it possible that I need to bump the Vcore (or lower the CPU speed) with more RAM installed, or is that a VTT voltage issue?

Any help is appreciated.

 

[Forceman]

Is it possible that processors require more voltage over time to hit the same clock speeds? Or that adding RAM can destabilize an otherwise ok system? Is it possible that I need to bump the Vcore (or lower the CPU speed) with more RAM installed, or is that a VTT voltage issue?

Any help is appreciated.

Yes, it is possible that processors require more voltage over time to maintain the same overclock. I think it is less of a factor now than it used to be, but it can happen. Adding RAM can definately cause instability in an overclocked system. The additional RAM increases the stress on the memory controller and could require more voltage to achieve the same speed - I'll caveat that by saying I'm not sure what happens in the i7/5/3 world since I've never overclocked any of them. Or, it could just be that one of the RAM modules isn't capable of the higher speeds.

 

[spotpuff]

Yes, it is possible that processors require more voltage over time to maintain the same overclock. I think it is less of a factor now than it used to be, but it can happen. Adding RAM can definately cause instability in an overclocked system. The additional RAM increases the stress on the memory controller and could require more voltage to achieve the same speed - I'll caveat that by saying I'm not sure what happens in the i7/5/3 world since I've never overclocked any of them. Or, it could just be that one of the RAM modules isn't capable of the higher speeds.

Hmm, I figured that might be the case since the IMC is, well, integrated now. The RAM is actually 1600MHz 8-8-8-24 RAM and it'd be running at 170x8 or 1360MHz, so it's actually underclocked and shouldn't have any problems

I asked someone at home to check for me and the system has passed over 15 runs of memtest now without issue; maybe loading the defaults fixed the issue? (I didn't do this after the previous BIOS flash... bad, I know, but why is this necessary?) In any case I will attempt to isolate the issue more now that it seems like the RAM is not the culprit.

On a side note, has anyone gotten the 21x multiplier to reliably stick on this board w/ the f13 bios? I had it at f5 but not with the f13 BIOS... why do they do that :T

 

[spotpuff]

Board passed 5 runs of burntest (max) at 180 bclk (6x RAM, 15x CPU) so it's probably not "just" the bclk that was causing issues.

I'll keep testing and tweaking I guess until I figure it out but overclocking non-methodically the first time sort of makes it hard to pin down problems. I'll test RAM next then CPU I guess.

 

[ghost6303]

packing more high frequency RAM modules into the same space creates more noise and cross-talk, regardless if you have the most stable voltage regulator in the world. sometimes an increase in volts helps, but not necessarily.

 

[drescherjm]

Can adding more RAM destabilize a system?

Yes especially if you populate all ram slots instead of removing the old ram. And also if you are overclocking the memory controller. Remember that 1333 MHz is the highest official supported ram frequency for AMD and Intel with Intel 1366 chips supporting 1066 MHz max.

 

[drescherjm]

Intel i5-750 @ 170x20 @ 1.125Vcore, 8GB DDR2 OCZ Gold 8-8-8-24, Gigabyte P55A-UD4P, Sapphire 5850

DDR2 in a P55 board? I do not think that is possible.

 

[Comte]

I recently added another 4GB of RAM to my motherboard, so now I'm running 4x2GB of DDR3.

I started getting random BSODs and crashes, but it's inconsistent. After running some tests in Windows at my OC'd settings using Linx, Prime95 and a memory testing program, it seems like there's an issue with some of the RAM (but not all of it); 3 prime threads will keep running but one will crash right away and the RAM checker only errors if it includes a certain portion of the RAM.

Decided to try Memtest86+ and on reboot it started errorring immediately. OK, that looks bad, so I reset the defaults in the BIOS and ran memtest again, and it started throwing errors again. I was playing around in memtest's settings to try the "probe" setting under how much memory to test, and the screen output became garbled and the computer seemed to hardlock.

Upon reboot (still at stock settings) memtest ran for 3 hours without any incidents. I'm still running it now (while I'm at work) so by the time I get home it will have been running for 14 hours or so.

Previous to this I had been running at 3.5GHz at stock vcore/vtt/everything (RAM @ 1.66v) but after a few months it appeared that speed was unstable. I updated to a new BIOS but I could no longer use the 21x multiplier, but I settled on 20x170 to hit 3.4 on stock and everything seemed fine (this is with the new RAM); linx for 10 minutes ran fine, etc.

Is it possible that processors require more voltage over time to hit the same clock speeds? Or that adding RAM can destabilize an otherwise ok system? Is it possible that I need to bump the Vcore (or lower the CPU speed) with more RAM installed, or is that a VTT voltage issue?

Any help is appreciated.

Dude, reduce to stock, run system. No errors? Great. Errors? Remove all but one stick of memory, run memtest. No errors? Great. Replace it with another one and test. Errors? You might have found your problem. repeat until you find the memory errors. Everything check out well? Then you start doing advanced troubleshooting. Voltages and what not...

But you must limit the variables to find the problem. and that means everything is suspect, even your previously stable overclock.

 

[spotpuff]

Dude, reduce to stock, run system. No errors? Great. Errors? Remove all but one stick of memory, run memtest. No errors? Great. Replace it with another one and test. Errors? You might have found your problem. repeat until you find the memory errors. Everything check out well? Then you start doing advanced troubleshooting. Voltages and what not...

But you must limit the variables to find the problem. and that means everything is suspect, even your previously stable overclock.

Yeah I did reset to "optimized defaults"... I have no clue why memtest was throwing errors.

I've tested the bclk at 160, 170, 180, 190 and now 200 (memory at x6, CPU at x15) and it's passed 1 hour of linx now at 200 bclk with no additional voltage to the VTT or Vcore.

Noooo clue what was going on before; next up is clocking the RAM a bit higher until it's around 1600MHz, then the CPU.

Thanks for all the help ; still trying to figure out if x20 multiplier will cause problems because it's "even" and maybe all this stuff is going right because I'm on a x15 "odd" multiplier.

 

[valtopps]

when ever installing all 4 or 6 slots with memory i like to match them exactly and you need to increase the nb voltage to stable them.