Memory Problem

[Basilisk]

'PBuilt a computer for a friend several months back, but it's been a nightmare ever since: 'crashes intermittently, even under next-to-no load (overnight w/ no user tasks running).

RAM has appeared to be the problem (beep code on booting at times), so I've gradually increased the voltage. Even at 2.4V on the RAM, it sporadically refused to boot, or crashed. I've just booted at 2.5v for the first time, but can't yet say that will make it stable.

Details:
H/W---
- Biostar TF7050-M2, X2 5000+, various PSUs (currently Antec Earthworks 380W)
- Samsung HD403LJ, Wireless card, using IGP, unused LightScribe DVD.
- 4X1GB Geil GX22GB8500PDC / pn105ST06CL7813
- Vista HP, case in a cool room w/ plenty of air
- no RAM OC (and little on CPU as it's an AMD!)

- System happily ran 2xPrime95 TortureTest for 36 hours w/ Ballistix Tracer 8500 @ 2.2v
- Reducing the number of RAM sticks and varying them and the slots they were put in had no effect.

The RAM is labeled as "2.1-2.4v"; I'm not happy running it at 2.5v [board's top-end] to check stability, but I'm trying it. I expected it to run stable at 2.1v. lol.

Q#1: Is Geil a problematic brand? Or is this product from Geil troubled?
Q#2: Is this a troubled board/chipset? [nv7050PV/630a]
Q#3: Is 2.5v a risk if it turns out to be stable? [RAM has classic heat-spreader/cover]
Q#4: Any next steps to recommend?

Any info from the Cognoscenti would be appreciated. Thanks...

 

[error8]

Pumping 2,5 volts in ddr 2 will kill it very quick, with or without heat-spreader, so that is not an option. Have you tried running memtest? Either your motherboard doesn't like that particular Geil ram, or it's just defective memory. Geil doesn't really make good quality ram. I also had problems with it some time ago.

 

[imported_Scoop]

You may need to increase the northbridge voltage with 4 sticks. Increasing memory voltage isn't the solution. Though if there was stability problems with one or two modules, then it probably won't help.

 

[Basilisk]

Originally posted by: error8
Pumping 2,5 volts in ddr 2 will kill it very quick, with or without heat-spreader, so that is not an option. Have you tried running memtest? Either your motherboard doesn't like that particular Geil ram, or it's just defective memory. Geil doesn't really make good quality ram. I also had problems with it some time ago.

Gack! Well, I hope it's not TOO quickly as I just left the house for two days with Prime95 running (2 P95 TortureTests: one RAM intensive, one other the general TT).

Thanks for the reminder about Memtest: haven't used it for years.

'Wish I'd known Geil wasn't well-respected when I ordered the RAM! Oh, the pain of knowledge. LOL

 

[Basilisk]

Originally posted by: Scoop
You may need to increase the northbridge voltage with 4 sticks. Increasing memory voltage isn't the solution. Though if there was stability problems with one or two modules, then it probably won't help.

Yup, problems with 1,2 or 4 sticks. And after I bumped the NBv, yesterday.

 

[n7]

2.4+V?

That's killer time...

Geil rates a lot of their kits with insanely high vdimm, which i don't understand, since unless they possess magical chips, anything 2.3+V will eventually cause failure from what i've seen.

Run Memtest86+ & then as many instances of HCI Memtest in Windows as is needed to stress all 4 GB & see what happens.

You say nothing about what speeds or timings you're running the RAM at, but mention you are OCing the CPU yet not the RAM.
It doesn't really work that way.
If you raise the HTT, that increases the speed of the RAM based on what ratios it's set to use, which would mean you've either underclocked the RAM originally or OCed it, unless you're got an unlinked ratio option in the bios.

Maybe post your CPU-Z tabs for CPU & RAM via a screenie?

 

[Mondoman]

It's not really that Geil is bad -- they've just joined the bandwagon of RAM manufacturers selling higher and higher overclocked RAM by boosting the voltage in their "specs." I certainly wouldn't trust any RAM with a spec voltage higher than 2.2V.

 

[Basilisk]

Again, the RAM had "2.1-2.4V" on its label, so I wasn't too worried for a test at 2.5V.

I won't be back to my house (if the RAM hasn't burnt it to the ground) until tomorrow. Then I'll try the recommendations above.


Regarding current clocks: (a) I only set the CPU +5% on the last day as I was getting bored; (b) when I set the CPU up 5%, I was presuming the RAM was decoupled, but I was harried as I left and need to check that again. All the recent failures (through 2.4V) were with CPU timings at default.


Yeah, I think Geil was fudging their RAM specs in claiming it is 8500 -- but I didn't think that at the time I purchased them. I naively thought "2.1-2.4V" meant it would be clean at 8500 with 2.1, and you could push it up to 2.4 in OC'ing it. Well, I emailed their support staff Friday; perhaps they'll have something useful to contribute.


Thanks to all for your comments.