How do you know if its NB, ram or cpu?

cje

cje

Associate
Joined
3 Jan 2009
Posts
786
Sorry to bother again.

I just don't know what various crashes mean. Is there a way to tell what causes a crash, by how how it crashed?

For example, I've heard that a restart is cpu problem. And that RAM/ NB is a BSOD with no restart or even a freeze.

I've found that bad ram overclocks nearly always end up either in a freeze, or a bsod that won't restart.

Any ideas? Thanks.:cool:
 
Hey cje,

the starting point for me personally when overclocking is a *stock* system. Once the PC has passed a few days of general use and stress testing then you have a stable base to launch your overclocking madness!

There is no right or wrong way to overclock but we all have out personal styles. I prefere to perform *isolated* overclocking tests where you apply pressure to each single component one at a time then you pretty much know where you stand when something fails.

You need a nice bunch of diagnostics and know how to use them, also a sound knowledge of the various bits and how they all relate to each other is helpful too! :)

A good method with some know how, the right testing tools, lots of free time and a wodge of patience will see you good! :cool:
 
Cheers for the reply Big Wayne, much appreciated.
Yeah, I've got my Q6600 totally stable at 3ghz with stock volts of 1.3v. Everything else is on AUTO though, so I guess I need to find some patience... By totally stable I gave it 24 hours prime smallFFTs. I'm ever changing my cooling solution, and the max it got to was 40 degrees on load, so happy happy with this clock at the moment, but need more though, its like a drug.

But is there a certain way to tell what's failing? I tried clocking this cpu higher for absoloutely ages, with no success. I could get it stable 12 hours prime95, but then as I'd start using normally for a few hours it'd BSOD.

I'm working around settings given to me by AndyOCUk as he clocked a near identical system to mine bar the GPU. So, I guess I need to find some patience.

But is there a way to know if it's one component more than the other? Say I'd be stable in prime and orthos, but a BSOD but no restart, that may indicate ram or nb right?

I know that an undervolted cpu will always give me restart, it might bsod but restart. A ram or mobo error will bsod and no restart. Does this make sense to anyone?
 
Well the first thing you can do is run memtest86 all night while you sleep to see if your RAM pops up any errors, which is where my BSODs were coming from. If you get errors then you need to find out which stick/sticks it is and RMA and/or replace. That would be my first option anyway. If you don't get errors then move on to the next tests.

Link to the proggy is in the "Useful Overclocking Applications" thread at the top.
 
Yeah, got Memtest already. But it will pass memtest and prime. Maybe I haven't tested them long enough. That did use to be the issue, as one stick of ram was indeed bad, but I still cant get it stable above 3ghz, even with good ram.
 
I've got my Q6600 totally stable at 3ghz with stock volts of 1.3v. Everything else is on AUTO though, so I guess I need to find some patience... By totally stable I gave it 24 hours prime smallFFTs.
Heh . . 24 hour Small FFTs primes! That's pretty extreme! :)

It's pretty common knowledge that Prime Small FFTs is a very rough guide to CPU stability only, that test puts a lot of pressure on your chip and pretty much bypasses the rest of the major system components. A system that is Prime Small FFTs stable can crash n burn very quickly once you play a game or perform some NB/Mem intensive tasks!

I've recently discovered that Prime Small FFTs is actually not 100% when performing *isolated* CPU testing and can give a false thumbs up but the processor will still fail due to undervolting/overclocking. In the scenario I encountered the fix for my random rebooting was to adjust the vCore up a few micro notches.

So basically don't rely on Prime Small FFTs giving you the all clear and instead use it as a real rough guide, even though it runs for 12-24 hours your chip may still need a little extra vCore

I'm ever changing my cooling solution, and the max it got to was 40 degrees on load, so happy happy with this clock at the moment, but need more though, its like a drug.
When you say you "need more" are you talking about more MHz or more cOOling? :D

If it's the former then you are showing signs of overclocking addiction, if it's the latter then you are showing symptoms of an overcOOling addiction heh! :p

It's deffo a good standard practice to keep your hardware nice and cool but you have to watch out for cooling for coolings sake, the reason you want to cool is to allow your hardware to be overclocked and run faster and in some respects to extend its lifespan! :cool:

But is there a certain way to tell what's failing? I tried clocking this cpu higher for absoloutely ages, with no success. I could get it stable 12 hours prime95, but then as I'd start using normally for a few hours it'd BSOD
Well you know what they say, if at first you don't succeed, try try again! . . . .also knowledge is power . . . .and not forgetting all good things come to those that wait! :o

There is a slim chance you have a Mc pOOp CPU but mores the likely you are doing something wrong/making a basic nOObilla mistake, without knowing more about your set-up and seeing some CPU-z screenies it's hard to help further . . .

regarding the BSOD there is every chance that even though Prime Small FFTs led you to believe your processor was stable at the XYZvCore you set . . your chip in fact needs a *little* extra voltage to run fully Real-World stable i.e not BSOD while you are inspecting some interesting pr0n! :eek:

But is there a way to know if it's one component more than the other? Say I'd be stable in prime and orthos, but a BSOD but no restart, that may indicate ram or nb right?
Yup that could be NB or RAM or some other aspect of the set-up like weak PSU, bad drivers, spilled kebab, etc etc, as mentioned above it could even be the CPU is thirsty for vCore but your not giving it because Prime Small FFTs misleads you! :(

I know that an undervolted cpu will always give me restart, it might bsod but restart. A ram or mobo error will bsod and no restart. Does this make sense to anyone?
It totally makes sense but there is no simple/quick answer.

The only way I can have any joy is too take things at my pace (i'm very slow but very methodical). If you read a lot of posts here you can see there are others who literally bang a system together and overclock it to the max straight away . . . ala "Straight in at 4.5GHz!" :o

Fair enough and they may just be naturally gifted and know a lot but most of the time you don't actually get to see any stability shots to confirm their overclock so who is too know if the system is good and robust.

Best to take things at your own pace and get to understand the patterns and symptoms of a stressed out PC.

I'm finding these two tools the most useful for my testing:

IBT (INTEL Burn Test)

Memtest86+ V2.11

rthdribl v.1.2

prime95 25.8
 
Yeah, got Memtest already. But it will pass memtest and prime. Maybe I haven't tested them long enough.m.
How long did you run memtest, my errors only started showing up after just over an hour. Also what are you temps like?
 
Last edited:
Ran memtest for 4 hours. No errors, but day to day activity always resulted in a random blue screen. As for making noobzilla mistakes, I have made so many. :D Like setting everything to a known stable 3ghz overclock, and didn't notice the multiplier was on 6. And tried raising fsb back up again. Oh, the memories. But from what I've heard the Striker II Extreme is a right pain to get stable.

My PSU should be just fine, it's a Zalman 1000 watt.

I think it's just the board being stroppy. I undervolted my Q6600 down to 1.2125v and it ran fine 5 hours prime95 at stock speed. So I guess my cpu is ok? It's VID is 1.3, so I guess my cpu is capable of more than 3ghz...

EDIT - Here are my settings for my 3ghz overclock, (the only thing I've changed is vcore and fsb) everything else is on auto:

n.b. I give BIOS values, and actual. The Striker II extreme overvolts everything.

Vcore: BIOS: 1.275v Actual: 1.33v vdrop: 1.328v vdroop:1.280v
RAM: 1.66v
VTT: 1.42v
PLL: 1.52v
NB : 1.33v
SB: 1.49v


VTT seems a bit high on auto?
 
Last edited:
Yeah, as we all know, leaving voltages on auto do tend to overvolt. I know that the CPU will run ok now with 1.275v in BIOS. But due to the striker overvolting, its actually getting 1.32v. Now I know the clock is stable, I'll work to see if the rest of these voltages can be lowered from their high auto settings :D.
 
Back
Top Bottom