• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Problem: MB, CPU or memory?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

KrisMCool

Member
Joined
Jan 17, 2004
Location
Greensboro, NC
First, I apologize if this gets too long. I'm sure some of you have read about my problems that I initially thought were due to my new OCZ memory. I'm posting here because I now think all the problems I've been having are due to my motherboard. Move the thread if it would be better elsewhere.

Components:

ASUS P4C800 Deluxe Rev 1.03, BIOS 1014
P4 2.8C SL6WJ Costa Rica
P4 2.8E SL79K Phillippines
2x512 Mushkin PC3200 LvII
2x512 OCZ PC3700 Gold Rev 2

Problem: (instability defined as errors in Prime95 ver 23.8.1 Blend torture test and/or FAH - both produce errors @ same settings even when other tests pass)

System loses stability above FSB 237-238 @ 5:4 & FSB 223-224 @ 1:1 regardless of CPU used, memory set used, dimm slots used, vcore (up to 1.65v 2.8C, up to 1.45v 2.8E), vdimm (2.85v max setting, not volt modded), vAGP (up to 1.7v) and PAT settings. Both CPUs run up to FSB 250 / 3.5GHz, but are very unstable. Mushkins start producing errors in Memtest86+ @ FSB 225, 2-3-2-6-8, PAT enabled - looser timings and PAT disabled had no effect. OCZs produce errors in Memtest86+ (tests 6 & 10) @ FSB 233, 2.5-3-3-7-4 (rated speed / timings), PAT disabled.

My guess: Because 2 CPUs & 2 sets of high quality memory fail at the same settings (FSB / ratio), I would guess this is MB related. Even though I'm on air, temps haven't been a problem. 2.8C - no higher than 41.5°C (DigitalDoc5+) @ 100% load. 2.8E - avg 41°C @ 100+% load w/ spikes to 50°C (MBM - reports higher than DigitalDoc for 2.8E). Chipset - no higher than 38°C @ 100% load (DigDoc).

Agree or disagree? Any advice would be greatly appreciated.
 
Last edited:
Been using 1014 since it was out. Got same results with previous versions. Think reflashing and clearing CMOS would help?
 
KrisMCool said:
Been using 1014 since it was out. Got same results with previous versions. Think reflashing and clearing CMOS would help?

I might help...but I kinda doubt it. You could have a funky mobo, stranger things have happened. Only way to tell is try your stuff on a different one. Got any friends with a P4 rig you can try your parts on?
 
I wish I did, but no. Already looking into another mobo. Getting another mobo wouldn't bother me (much) if there were some way I could be sure that's where the problem lies. I'm not expecting any magic answer, just trying to narrow it down as much as possible.
 
I'm also definitely going to go to water for this Prescott. Gonna build my own setup with White Water blocks, since plumbin's what I do best!
 
I agree that testing components on a friend's mobo is the way to go... it gets expensive (believe me, I know) to start buying a bunch of >$200 parts tearing your hair out trying to figure out wtf is wrong..
banghead.gif


one question, have you disabled "Spread Spectrum" in the BIOS? From what I know, this option makes small fluctuations in the FSB to keep high-fsb mobos from interfering (or being interfered with) by Radio waves... BUT it supposedly reduces stability when you start OC'ing. I'm sure you know your stuff from reading your other posts so you've prolly checked it already but, just makin sure.

I assume you checked that the heatsink was making good contact with your Northbridge as well?:rolleyes:
 
In my case, I didnt touch "Spread Spectrum" so i guess its on default, yet its perfectly stable :)
 
well from what I know, you will get better performance with it OFF, and perhaps even a bit more headroom to crank your fsb up
 
Here are some of the other BIOS settings I'm using. If you see something wrong or forgotten, let me know.

Spread Spectrum: Disabled (always)
ICH Delayed Transaction: Enabled (always - no difference noted disabled)
AGP/PCI Frequency: 66.66/33.33 (always)
Performance Mode: Auto (presently)
Performance Acceleration Mode: Enabled (presently)
Plug and Play O/S: No (always)
PCI Latency Timer: 64 (always - default)
PCI IDE BusMaster: Enabled (always)
Legacy USB Support: Disabled (always - no probs noted when Auto)
ACPI 2.0 Support: Yes
ACPI APIC Support: Enabled
BIOS -> AML ACPI Table: Enabled
Power Management / APM: Enabled
32Bit Data Transfer: Enabled (presently - no difference noted)
 
KrisMCool said:
Here are some of the other BIOS settings I'm using. If you see something wrong or forgotten, let me know.

looks perfect to me! same settings here... the PCI latency / ICH delayed xaction are 'up for debate' as some do better with them off or lower, but I find my RAID performance goes down when I set it to 32 or disable ICH delay....
 
I have tried several Northbridge cooling solutions with only small differences in temperature noted. ThermalTake Tiger 1: poor base finish - had to lap, good retention, quiet fan, better than passive, inexpensive. Abit's current NB cooler: annoying high-pitch fan, interferes with larger Thermalright HSs, poor retention, better HS than TT but only 1°F cooler, very inexpensive. Swiftech MCX159-P: high-quality finish, very quiet fan (but no RPM measurement :( ), 2-4°F cooler than TT, poor retention, expensive.

Currently using Swifty with AS5. The spring clips are annoying, but worse I don't think they allow enough pressure to be applied to the chipset. 4 clips is much tighter than 2, but I'd prefer screw mounts. The largest improvement in NB cooling came from installing a side-door mounted fan, meant to cool the AGP card but it really helped NB temps under load. I might modify the Swifty to use all 4 clips with another spring. Current temps run 98-99°F at full load. While I'd like to improve those numbers, I don't think it's the problem.
 
KrisMCool said:
I might modify the Swifty to use all 4 clips with another spring.

well the current swifty now ships with all 4 spring clips. I'm sure you could call them and ask for the 2nd clip. BUT I agree, doesn't sound like your NB is the problem...

I read your 1st post again and something that's steering me AWAY from it being your Mobo (aside from Asus' generally excellent reliability) is that you said:

System loses stability above FSB 237-238 @ 5:4 & FSB 223-224 @ 1:1 regardless of CPU used, memory set used, dimm slots used, vcore

now if it was your mobo, the FSB points at which you lose stability should be a little closer together, dont you think? The fact that you can run higher fsb @ 5:4 points more towards a faulty DIMM. Have you tried running with only 1 stick? Or just get some entirely new sticks.... or borrow,etc... just to test. I know its a long shot since you've tried 2 different sets of DIMMs but it's always a possibility. I will admit that yours is a very strange problem........

anyway, might be time to take advantage of those "lifetime warranties":rolleyes:
 
KrisMCool said:
Nice new avatar, BTW. Know any good places to look for avatars?

THANKS!:p I am always a big fan of
google.gif


i used the 'image search' and found mine in like 2 mins... a couple more mins in photoshop and BAM... instant avatar!:cool:
 
Memory is what I originally thought it was, after buying the OCZs. But the FSB / ratio limits were the same for both sets of memory. The OCZs only produced errors (at rated speed & timings) in Memtest with several hours of the extended test, with most of the errors in test 10. Don't know what test 10 errors prove. And the Mushkins are rockin' (see sig) without a vdimm mod. I was told that running 1:1 at higher FSBs stresses the memory controller (mobo) much more than 5:4 at the same FSB. That's why I didn't think the differences pointed to the memory. I almost RMA'd the OCZs. I will RMA, but only if I can prove it's not something else first. Getting another set of memory is a long shot, and single sticks didn't allow much higher.

My 3.3v rail might be a problem. Not the PSU (see sig) - it's dead on (voltmeter tested). But how the mobo handles and distributes that rail. All software measurements show low voltage (3.25v avg w/ numerous drops to 3v). Had the same issues with my previous PSU (PC Power & Cooling Silencer 400), so the PSUs aren't the problem.

Already ordered a P4C800-E so I can get a definitive answer. Expensive, but I wanted a second rig and this was my excuse. ;)

Thanks for the update on the Swifty - I'll look into that.
 
I see what you mean, yes you are right about running 1:1, and getting 225mhz out of the Mushkins w/o a VDIMM mod is not too shabby actually!! I think most peeps are seeing those sticks top out around ~230Mhz (DDR460) and that is WITH a vdimm boost, so I guess you are right: your RAM is probably fine

Well do let us know how it goes with the P4C800-E...
(that's what I'm running-- Damn near perfect mobo actually, if they would have just put some active cooling on the NB... :p )
 
Back