Stability Issues

[asmolow]

Setup:

Gigabyte PA-35-DS3L Rev 2
Q6600
4GB OCZ Platinum (4x1)
OCZ 700w Gamexstream


I am having a terrible time trying to get my overclock stable. I've played around with overclocking from 2.8-3.2.

Prime95 usually gives me an error after 2-4 hours along with random BSOD's now matter what I set my FSB and multiplier too. I have increased voltage almost to the 1.5v limit while under load and I still have stability issues.

When I run memtest with all 4 sticks of ram in, I usually end up with one error after running it for hours, but when I test each stick individually I do not receive any errors after 50 passes. I have relaxed the times to 5-5-5-15 and increased voltage to 2.1 and 2.2 without any luck.

I am not sure what this points too, but my computer appears to become stable if I only have 2 sticks of ram in use. I am still in the process of testing and optimizing so it is mostly subjective, but crashes have become much less frequent.


Any ideas as to what is causing the unstability and how I can use all 4gb of ram?

 

[Evilonigiri]

If your ram is rated at 2.2v, you'll have to set it at 2.0v. Try tweaking the ram around a bit. The motherboard could be giving out more than it says it is.

 

[asmolow]

I will give it a try.

I tested all the sticks individually at 2.1 and they all passed as well as it being stable with just two sticks in use at 2.1. Would having all 4 in at the same make a difference?

 

[Evilonigiri]

I will give it a try.

I tested all the sticks individually at 2.1 and they all passed as well as it being stable with just two sticks in use at 2.1. Would having all 4 in at the same make a difference?

If it passed with 2 sticks at 2.1v, then if you have 4sticks it should be higher, like 2.2v. More sticks means more voltage, so try 2.3v?

It could be a defective slot. Have you tried all the ram slots using a single stick of ram?

 

[asmolow]

If it passed with 2 sticks at 2.1v, then if you have 4sticks it should be higher, like 2.2v. More sticks means more voltage, so try 2.3v?

It could be a defective slot. Have you tried all the ram slots using a single stick of ram?

I'm reluctant to go beyond 2.2 because it voids the warranty. I haven't tested each slot individually. I'm in the process of testing channel 1 (2 sticks dual channel) or whatever it would be right now, if it checks out i will switch over to channel 2.

 

[asmolow]

I just realized that 4gb of ram runs completely stable at stock settings. Is it possible to have a defective slot that only shows up when overclocking?

 

[Mondoman]

Of course.
However, rather than fiddling with the RAM voltage, you may need to boost the memory controller voltage (MCH/northbridge) a notch or so. Running 4 sticks of RAM on today's consumer boards that use "unbuffered" RAM is pushing the limits, since each additional stick plugged in means the control signals are split among more DIMMs, and are thus weaker for each individual DIMM. Boosting DIMM voltage even further will not help with this issue.

 

[asmolow]

Of course.
However, rather than fiddling with the RAM voltage, you may need to boost the memory controller voltage (MCH/northbridge) a notch or so. Running 4 sticks of RAM on today's consumer boards that use "unbuffered" RAM is pushing the limits, since each additional stick plugged in means the control signals are split among more DIMMs, and are thus weaker for each individual DIMM. Boosting DIMM voltage even further will not help with this issue.

I have already tried this without any luck. When I did boost the MCH voltage, I did not reduce the ram voltage, so would that cause the ram to receive too much voltage?

 

[Mondoman]

That's correct -- just leave the RAM voltage at whatever its manufacturer's spec voltage is.