Sign in with
Sign up | Sign in
Your question

Orthos fail................

Last response: in Overclocking
Share
May 18, 2007 12:09:40 PM

Hey people, wonderig if someone could help

I have a moderate overclock on my e4300 (2.4ghz) and I'm using Geil ultra ram (800mhz, 2 gig)

When i run orthos cpu stress test I have no problems and my cpu stays at a respectable 45 degrees. When I run the ram test I also have no problems. However, when I run the combined test, I get a red bar saying "paging" and then a critical error failure after only 4 seconds. This does not, however, cause my pc to crash, and in general use (including high definition gaming) I have no problems what so ever, and in 30+ hours of use (its new) haven't had a single crash.

any ideas what can cause this?

Thanks in advance
Sam

More about : orthos fail

May 18, 2007 12:58:17 PM

What do you mean by 'the RAM test', etc.... as far as I know the one the stresses RAM the most is the 'blend' test, which IS a combination test of CPU and RAM.... is this the one you mean that fails? What is your RAM voltage set at and what timings are you using?
May 18, 2007 1:00:55 PM

there is one that stresses memory - cant remember what it is called. I'm in work at the moment, will check when I get home. Ram is running at 2.1v and 4-4-4-4-12 (is that the right number of 4's?) as per the specs
Related resources
May 18, 2007 1:09:40 PM

There are 4:

CPU
CPU (gromacs core)
CPU & some RAM
CPU & RAM

Have you tried testing the RAM using memtest or some such?

One too many 4s there :) 

What motherboard have you got?
May 18, 2007 1:13:43 PM

haven't tried testing the RAM yet, kind of assumed that since my pc was doing everything I asked it too, including some pretty intense gaming, and doing it FAST, then my RAM was all good?

surely if my RAM was faulty I would be getting crashes in other places?


I'm using a DS3.


thanks for your help btw, much appreciated
May 18, 2007 1:27:33 PM

Well, possibly... RAM could be just a bit defective which might not affect apps at first until it worsened. On the other hand though, Orthos is quite extreme, so it's possible that it may be stable enough for every day use, although failing after 4s is pretty quick.

Have you set the voltage to the RAM and timings yourself? Know that there were some issues with that board on early BIOS revisions, but that was more to do with getting it to boot for the first time.

Run Memtest86 or Microsoft Memory Test Utility for about 15 minutes later and see if it comes up with anything... the Microsoft one tends to find stuff quicker so that'd be best really, memtest is supposed to be an overnight job :( 

http://www.softpedia.com/progDownload/Microsoft-Windows...

Always glad to help
May 18, 2007 2:37:25 PM

just remembered that it says something like "0.5 rounding error, was expecting 0.4" (or something almost kind of similar)

does anyone know of a list of orthos errors somehwhere on the net?


thanks again
Sam
May 18, 2007 2:52:21 PM

I think this is just a standard message that signifies that the calculation has not produced the result that was expected, meaning that the components are not functioning correctly. I get this message when either my CPU overclock is unstable, or my memory overclock is unstable, doesn't seem to matter which.
May 18, 2007 3:43:30 PM

Quote:
Hey people, wonderig if someone could help

I have a moderate overclock on my e4300 (2.4ghz) and I'm using Geil ultra ram (800mhz, 2 gig)

When i run orthos cpu stress test I have no problems and my cpu stays at a respectable 45 degrees. When I run the ram test I also have no problems. However, when I run the combined test, I get a red bar saying "paging" and then a critical error failure after only 4 seconds. This does not, however, cause my pc to crash, and in general use (including high definition gaming) I have no problems what so ever, and in 30+ hours of use (its new) haven't had a single crash.

any ideas what can cause this?

Thanks in advance
Sam


What are your CPU FSB settings, CPU voltage?

I have found on my system that running the memory unlinked, the memory FSB is most stable and makes Orthos (happy) at the speed of CPU FSB divided by 4, times 2.

Meaning the system FSB for my PC is 1800Mhz @ 3.6Ghz, divided by four is 450 times 2 = 900Mhz for my memory speed (Unlinked) of course.

If your system FSB is an odd number but your memory is set to 800Mhz, normally your BIOS will auto configure itself to the closest divider but I have noticed even 5Mhz difference will cause instability between the CPU FSB and the memory FSB. I have to speculate that it has to do with how the to buses synchronize data back and forth which leads to errors if the values are off.

And the CPU voltage.., your system may need a little bump in the voltage juice to help increase its stability, but in your case it sound like just small amount.

Anyways, good luck, maybe this helps.
May 18, 2007 4:22:08 PM

Quote:


And the CPU voltage.., your system may need a little bump in the voltage juice to help increase its stability, but in your case it sound like just small amount.



This shouldn't be a factor as he passes the CPU test, just not the blend.....
May 18, 2007 4:44:52 PM

Quote:


And the CPU voltage.., your system may need a little bump in the voltage juice to help increase its stability, but in your case it sound like just small amount.



This shouldn't be a factor as he passes the CPU test, just not the blend.....

maybe, maybe not, when overclocking, even mild OC, a little bit of extra juice won't hurt just as a test and you would be surprised how often that fixes, "weird" errors.

If it makes no difference it won't hurt and he can set it back.
May 18, 2007 5:39:43 PM

It might not be a weird error though...... if he's not putting enough volts through the RAM an increase on the vDIMM would be more likely to sort it, but lets see what he comes back with when he checks his settings...
May 18, 2007 6:40:53 PM

fsb settings 275 x9. cpu voltage is 1.325. looks like i'm failing the memory test too now :( 

have bumped the mem voltage up to 2.2 and am running it at stated timings of 4-4-4-12 - starting to think i may have a dogy ram stick

any suggestions?
May 18, 2007 8:36:36 PM

So you've only upped your FSB by 8? Try upping the CPU voltage a notch.... would be very surprised if that were the problem but worth a go. That RAM doesn't like extra voltage so I'd stick with 2.1v. Try running that memory tester I linked and see what it comes up with.
May 19, 2007 11:44:24 PM

only, i've worked out what the problem was, and its getting me a bit pissed off...... I realised that I was actually running a 4:5 fsb:ram ratio. when I switch it up to 1:1 I'm all good, currently running at 2.7ghz with an orthos load temp of about 48 degrees.

I've also passed orthos blend test for 40 mins - i know you guys like to leave it longer but the fact that its working that long shows me my system is loads more stable..

the only thing is though, at this speed my ram is effectively running at 600mhz, so although I'm gaining 40%+ on my CPU speed I'm losing 25% on my ram speed. Is the extra CPU performance worth it, or will it not become a viable trade off unless I clock my chip higher? does anyone have an idea why my ram wont run at 4:5, even though thats well within its stated speed?

thanks again guys
Sam
May 20, 2007 1:46:22 AM

Glad you got it sorted.... funnily enough was reading just today that that divider causes problems for some people... the extra CPU speed will more than make up for the RAM. If you are unhappy that your memory is running slower than stock you could loosen the timings to 5-5-5-15 and set the divider to 2:3. C2Ds like higher frequency as opposed to tight timings anyway.
!