Should I be worried about OCCT detecting errors?

xinaes · Post by **xinaes** » Sat Jun 11, 2011 9:36 am

I've just built a machine similar to the one I specced up in
viewtopic.php?f=23&t=62463
(slightly different RAM, HDD... maybe one or two other minor differences).

I've been running some stress tests, and most things I've tried run absolutely ok. However, OCCT does seem to produce a few errors. I'm almost inclined to ignore it, but that doesn't seem right...

I made a thread about it on bit-tech
http://forums.bit-tech.net/showthread.php?t=211199
but thought I'd post here too in case anyone can offer some advice.

Thanks

HFat · Post by **HFat** » Sat Jun 11, 2011 1:51 pm

Stability testing is not trivial. Some tests can reveal issues others will miss.

Don't assume the problem comes from the CPU (assuming there really is a problem) if it shows up at stock voltages and clocks. RAM problems are more common.

Any reason to trust that OCCT thing?

xinaes · Post by **xinaes** » Sat Jun 11, 2011 10:41 pm

Thanks for the reply. I suppose I don't have any particular reason to trust OCCT and on advice from bit-tech forum I'm going to go for a dose of LinX next, see how that goes down. I don't see why I particularly wouldn't trust it.

Of more particular relevance to this forum, I need to find out why my case fans are making so much noise, but that's for another thread...

b_rubenstein · Post by **b_rubenstein** » Mon Jun 13, 2011 6:56 am

OCCT is running Linpack and if it's indicating errors then your computer isn't stable when running that program. It may never throw an error in normal usage, or it may. If all you're doing is playing games, then it may not be an issue.

If any of the computers in my house (mine or my kids) can't run the OCCT stress test for an hour, the I change settings until it can.

xinaes · Post by **xinaes** » Mon Jun 13, 2011 7:32 am

b_rubenstein wrote:OCCT is running Linpack and if it's indicating errors then your computer isn't stable when running that program. It may never throw an error in normal usage, or it may. If all you're doing is playing games, then it may not be an issue.

If any of the computers in my house (mine or my kids) can't run the OCCT stress test for an hour, the I change settings until it can.

OCCT seems to have two CPU stress modes; OCCT & Linpack - perhaps the former also involves Linpack, I'm not sure. In any case, the worrying thing in my case is that I couldn't get it to complete a complete 1hr test under any settings that I've found.

I'll be using the machine for some gaming, mostly programming, quite a lot of photo editing, sound and graphics synthesis and increasingly much sound and video editing as well... so it's not like an error on the CPU is going to cause meltdown of any nuclear reactors, but obviously I'm trying to ensure that the machine is properly set up and stable.

Unfortunately, I have encountered a few more 'real' problems as well as the more 'theoretical' OCCT failures. Sometimes it seems to freeze before getting past BIOS, needing a hard reset perhaps two or three times in a row before working absolutely fine. I've also seen it freeze at the latter stages of booting windows (late enough that the mouse cursor is there and works for a time and then just appears frozen). I've also had two BSODs to date; once was while I was watching a YouTube video - I don't think much else significant was left running. Another was right at boot time.

None of these appear to be related to the computer being stressed, or even particularly correlated to overclocking or any other BIOS settings or really anything that I can think of. For example, I seem to see approximately the same frequency of CPU errors in OCCT with stock settings. Although I'm quite a technical user, this is the first time I've had a desktop of my own for many years, so I'm not really used to having much to tweak in the way of BIOS settings etc - and I'm definitely more of a software person.

At the moment everything is running very cool. Once I'm more satisfied that everything is stable I'll look at calming down the case fans a bit as they are pretty audible at the moment.

I think I should do some more specific RAM tests perhaps. Pretty sure it had a clean run from at least one such test, though.

Excuse my verbosity, and thanks for the input.

HFat · Post by **HFat** » Mon Jun 13, 2011 7:48 am

It sounds like all your problems happen while in the same OS install. You might want to try running another one to rule out software issues.

RAM is cheap. Even if you have no other computer with compatible RAM, you could purchase a small stick for testing. If you have two, you could remove one of the two.
Memtest is popular but I'm not sure it can catch every real-world problem. But if you haven't yet let your computer through a few hours of memtest, do so first thing!

The next thing to do if it doesn't seem to be a RAM problem would be to reset all BIOS settings and test your computer with a known-good OS (not the one on your hard drive which could carry unstable settings or malware) such as a Live CD.

You said you had no particular reason not to trust OCCT. Maybe you should be more careful. Obscure software, next to no documentation, no source available... it could easily carry a trojan or simply be crap.

xinaes · Post by **xinaes** » Mon Jun 13, 2011 8:34 am

HFat wrote:It sounds like all your problems happen while in the same OS install.

Apart from things like not getting past BIOS sometimes... (incidentally, I'm glad I could change the BIOS logo graphics on the motherboard - I'd be pretty fed up with the Gigabyte graphics by now otherwise!)

Memtest is a good call. If it does show up any problems then I'll try switching the RAM.

I have Ubuntu installed as well, which I really haven't done much with as yet. I might try running some tests there, or from a live cd as you say.

You said you had no particular reason not to trust OCCT. Maybe you should be more careful. Obscure software, next to no documentation, no source available... it could easily carry a trojan or simply be crap.

meh

... well, I'm fairly used to being careful on t'internet and am pretty sure my judgement wasn't too far off here, nothing flagged etc... admittedly, a colleague of mine who's a highly competent (by my standards

) programmer / biophysics researcher and generally pretty clued up recently contracted a virus from a website through an unpatched version of Flash... schoolboy error you may say, but it shows that one shouldn't be complacent.

xinaes · Post by **xinaes** » Mon Jun 13, 2011 11:15 am

Running Memtest now, it just hit an error after ~1:53. I guess I'll take a stick out and leave it running Memtest again while I go to the pub.

HFat · Post by **HFat** » Mon Jun 13, 2011 12:22 pm

Even if memtest didn't show any errors, I'd have tried switching the RAM.
Sometimes sticks that seem to work well in a computer behave oddly in another. I haven't always been able to explain to my satisfaction the RAM-related problems I've dealt with. I guess bad boards can make RAM look bad.

I don't know that the memtest error you got is proof positive that the error comes from the RAM but you'll know if you can find a stick that doesn't throw any errors.
It may be that all your sticks are bad, especially if you got identical ones from the same retailers.

A bad stick of RAM may not throw errors immediately but become a source of errors instability after a few months. So if you get rid of your problem by getting rid of a bad RAM stick, don't assume the other one is good. Test it again after a while, especially if you notice instability again.

xinaes wrote:contracted a virus from a website through an unpatched version of Flash

A fully patched version of Flash is unsafe in my book. It's almost guaranteed to be vulnerable to 0-days. I wouldn't browse any random website with Flash. Flash ain't open so, if I can help it, I never use Flash. But if you really want to use Flash, don't give it elevated priviledges!

xinaes · Post by **xinaes** » Mon Jun 13, 2011 3:12 pm

HFat wrote:A fully patched version of Flash is unsafe in my book. It's almost guaranteed to be vulnerable to 0-days. I wouldn't browse any random website with Flash. Flash ain't open so, if I can help it, I never use Flash. But if you really want to use Flash, don't give it elevated priviledges!

Indeed, it does seem hard to imagine how Flash could have had adequate priviledges to let in whatever it was that it did... not sure of the details.

Thanks for the other advice. Currently one stick of RAM has reached 3:37 without error. I'll leave it going while I sleep. I didn't notice all of the figures on the Memtest screen before, but this time I notice it's clocked at 535MHz (DDR3-1070), rather than ~666MHz which is what I'd expect IIRC.

ntavlas · Post by **ntavlas** » Mon Jun 13, 2011 5:14 pm

A friend's computer showed very similar symptoms, turned out it was the ram. It wasn`t faulty though, choosing more relaxed timings in the bios was enough to solve the problem.

xinaes · Post by **xinaes** » Tue Jun 14, 2011 1:18 am

Stick #1 ran Memtest for ~10h:30 with no errors. Switched to the other (in the same slot) this morning and it hit four errors in 1.5 minutes! Having said that, this was around the time that something in the house tripped a circuit breaker and I'm somewhat inclined to discount that result if it is not repeatable.

I'm repeating the test again, without having reseated or changed anything and it's been going for 34 minutes without error. edit: I let it get very close to the end of a complete run through and it was fine. I then accidentally hit escape on the wrong keyboard...

p.s. looking at the BIOS, I can't see why Memtest was reporting memory clocked at 535MHz.

Should I be worried about OCCT detecting errors?

Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?

Re: Should I be worried about OCCT detecting errors?