ORTHOS fail after 16 hours

[mesias]

Hi all

This is my first rig for overclocking so please bare with me. After numerous try and error, and a lot of reading i managed to reach some apparent stability on the board. These are my settings:

* All unused options in the BIOS are disabled
* SLI Ready Memory:Disable
* CPU Operating: Unlinked
* FSB: 1600 MHz
* Multiplier: x8
* MEM: 800 MHz
* N/B Strap CPU: PSB1333
* All PCIe: 100 MHz
* CPU Core: 1.415v
* DDR2 Voltage: 2.2v
* CPU VTT: 1.2v
* NB Voltage: 1.35v
* SB Voltage: 1.65v

After 16 hours of ORTHOS running, one of the cores returns an error and the test is stopped. It is very unlikely that a regular user, or even an extreme gamer, would keep a system running at full load this long. Can this result be consider as a positive sign of stability for the overclocked settings? For how long a systems is supposed to run at full load to be considered stable?

[freakBYnature]

Hi all

This is my first rig for overclocking so please bare with me. After numerous try and error, and a lot of reading i managed to reach some apparent stability on the board. These are my settings:

* All unused options in the BIOS are disabled
* SLI Ready Memory:Disable
* CPU Operating: Unlinked
* FSB: 1600 MHz
* Multiplier: x8
* MEM: 800 MHz
* N/B Strap CPU: PSB1333
* All PCIe: 100 MHz
* CPU Core: 1.415v
* DDR2 Voltage: 2.2v
* CPU VTT: 1.2v
* NB Voltage: 1.35v
* SB Voltage: 1.65v

After 16 hours of ORTHOS running, one of the cores returns an error and the test is stopped. It is very unlikely that a regular user, or even an extreme gamer, would keep a system running at full load this long. Can this result be consider as a positive sign of stability for the overclocked settings? For how long a systems is supposed to run at full load to be considered stable?

Not knowing much about the Intel apart from a friend of mine who OC'd his E4300 to 3.3-3.4 GHz... but I can without knowing Intel say that you prabably have to raise you voltage to cpu just a little noth if it takes all 16 hours for orthos to stop. My friend is using the Zalman9500 cooler aswell, he can boost it up to 1.6V without issuses, but then he also pencilmodded his board in order to prevent any drops in voltage. Also check your ram, is it going to high for what it can take, or are you using a divider that sets it right? Northbridge controls RAM on Intel, so if you are going for an higher FSB than RAM can take I also would suggest to try and up the voltage to Northbridge to test if stability is better there. And most important; try one voltage at a time, first the vcore - > orthos ... still fails? Leave the vcore at where it fails and try to up your Northbridge... and even more important; small increasments! Hope your are somewhat helped by this :rolleyes:

With full stability Orthos should run forever without an issue, to answer your very last question. Orthos is just a mathematical software that uses you cpu to calculate every math that the cpu is able to calculate, which an everyday use won't show.

[mesias]

From this point any slight change in the CPU voltage is giving me an F5 error and the system either halt or reboot. This memory can reach 500MHz at stock voltage but i haven't been able to raise the FSB to meet with the RAM.

[ineedaname]

Are you using small fft or large fft or blend?

That would make a difference.

Are you saying that more voltage is causing the system to not boot?

There could be many reasons why your computer is doing that. 16 hours is pretty stable but if you only did smal fft I would do some blend also.

the only real problem is that right now its hard 2 tell if it was the cpu or ram. However if more voltage is causing ur cpu 2 hang then that's an entirely different problem.

You could try raising your CPU VTT or even ur NB voltage cuz they're both relatively low.

[freakBYnature]

I was wondering myself.... small fft tests cpu, but blend actually tests more ram... if you use custom you can set test to run 1 minute each fft, so you'll see faster where it will fail... rather than 15 minutes wich is standard.

[mesias]

i've been using blend all along. What could be the problem with voltage and the CPU? For sure, from this point any higher voltage caused me problems

[Emerica]

Personally, if your computer is 16 hours Orthos stable and you don't expeirience any crashing while usually use of your computer than your fine.

[[XC] riptide]

Depends really on your use pal! if you do a little gaming, etc I think you're safe. You may get the odd 'freak' crash during high load but I'd say your safe.

On the other hand, if You run DC projects the rig ideally HAS to be stable 24/7 at full load. Thats days/weeks on end @ 100%. So again its all about your own usage.

[mesias]

I was wondering myself.... small fft tests cpu, but blend actually tests more ram... if you use custom you can set test to run 1 minute each fft, so you'll see faster where it will fail... rather than 15 minutes wich is standard.

Is there any way to concentrate the tests in the CPU or the memory to isolate what's causing the problem? would mind explaining more?

[freakBYnature]

Is there any way to concentrate the tests in the CPU or the memory to isolate what's causing the problem? would mind explaining more?

Again, small FFT's will ONLY test CPU, Blend will Mostly test RAM, almost NO CPU, but MEMTEST86 will ONLY test RAM :)

EDIT: Memtest86 you will need to download a cd-image, burn it to a CD and run in boot before XP.

[Mad_Man]

What could be the problem with voltage and the CPU? For sure, from this point any higher voltage caused me problems
HEAT

Can you please post the full load temps? from both cores. If higher Vcore = problems then CAN and possibly IS temp related.
Just my 0.5

[laragirl83]

Again, small FFT's will ONLY test CPU, Blend will Mostly test RAM, almost NO CPU, but MEMTEST86 will ONLY test RAM :)

EDIT: Memtest86 you will need to download a cd-image, burn it to a CD and run in boot before XP.

there is another topic here, which is the best tool for testing RAM stability!
we discussed lot of thing, but the point is, use GoldMemory v1.70 or newer instead of memtest86+...:fact:

[dinos22]

memtest is not used for full stability :confused: but is an excellent program to determine fairly quickly where you're at with your OCs :)

i use memtest + Spi and couple of hours of orthos if i can be bothered :)

i think the main idea of stability is when you use a system for a few months and don't experience any major issues :)

[theteamaqua]

i say PSU or board vdroop .. if there is even a moment ur CPU not getting enough vcore .. say u need 1.38v for 3.2GHz .. if there is a moment vcore drops below that .. its very possible that it will fail prime

[theteamaqua]

memtest is not used for full stability :confused: but is an excellent program to determine fairly quickly where you're at with your OCs :)

i use memtest + Spi and couple of hours of orthos if i can be bothered :)

i think the main idea of stability is when you use a system for a few months and don't experience any major issues :)

i agree .to be honest dual/triple/quad pi 32m stable is more than enuff for me ...

[freakBYnature]

there is another topic here, which is the best tool for testing RAM stability!
we discussed lot of thing, but the point is, use GoldMemory v1.70 or newer instead of memtest86+...:fact:

ok then :)

[ziddey]

something isn't right if setting higher vcore results in problems. that could very well be the root of issues

[rogard]

i would increase the NB volts, a E6600 should not require 1.41 to get to 3.2.

Increase the NB then, if it is stable drop the vcore back to say 1.35ish