can't get stable in Prime95 for 10+ hours

[FSBxtreme]

Hi all,

I have been testing stability for my new memory setup for a week now and I have to say I'm very tired and dissapointed. As you can see that prime95 only fails the between 10-24 hrs stress. I just purchased my PDP Patriot XBL PC3200 2x256MB kit and running it in slot 1 and slot 3 on my Abit NF7-S v2.0 mobo with Mantaray XT(reflashed from Merlin taipan because of this unstability) After the flash, it failed somewhere around 12 hours on dual channel. So to rule out the memory fault, I ran each module on slot 3 and 1 modules passes prime95 for more than 24hrs straight. The other module failed prime around 15-16hrs. So with this module, I double check it by run it through Memtest86+ default test, not a single error has been found for 24+ hrs.

So can I now rule out my memory? or should I memtest it in dual-channel first? Or your suggestion what to do next? Thanks.

Here's the list of component:

AMD XP-M 2200+ 35W
Abit NF7-S v2.0 with Mantaray XT cpc on
Patriot PC3200 256x2 with XBL technology (slot 1 & 3)
ATI Radeon 9500pro with Warp bios but not overclocking
Seagate 80GB 8mb HD
PC Power and Cooling 310W Silent series
Swiftech MC462+ with Thermaltake SilentCat
Northbridge and Southbridge are all passive-cooled.
2x120mm and 1x90mm with SuperFlower Fan Master

Current setting for each components:
-CPU : 200MHz FSB x 8.5 ; 1.475V (BIOS read 1.46 and 1.45V, AbitEQ reads 1.44 and 1.46V)
-Memory: 200Mhz FSB synchronous; 2.6V (Bios reads 2.61V, AbitEQ reads 2.61)
- VideoCard: no change, stock
-Temperature(idle/load): 38C/42C

 

[c627627]

Is that a mobile 2200+, those overclock as bad as most dekstops. Only 2400+, 2500+ and 2600+ mobiles overclock great...

Sorry for the off topic info, it's for other users reading the forum.

Anyway, bump the Vdimm RAM voltage to the highest setting, and increase Vcore voltage if your cooling can handle it, then report back if you still can't pass prime...

 

[Miguita]

Prime95 isn't completely reliable. One time your rig can run for hours and hours, but the next time you test, it lasts maybe just a couple of hours. I usually run the blend-test at least two times on the same settings (2x12 hours).

 

[Ad Rock]

Yep when in doubt, just crank the voltage . Also I wouldnt worry about being Prime unstable if you are stable in your other stress tests. Prime is a nice indication but it not everything.

 

[FSBxtreme]

Increase voltage for both components didn't help at all. Yesterday, I tested my memory modules in dual-channel with memtest86 and found thousand of errors within 30mins of looping Test #5. Voltage did not help in this case but the timing does. I change cas from 2 to 2.5 and the memtest85 stops detecting error after 60 passes of looping Test#5.

SInce my ram is rated 2-2-2-5 1T at 200Mhz FSB, and I can only run it at 2.5-2-2-5 1T at 200Mhz.

Should I consider RMA it or it's just AMD that likes 2.5 cas?