Sign in with
Sign up | Sign in
Your question

CPU fail orthos, then seems ok?

Last response: in CPUs
Share
July 17, 2009 2:06:02 AM

what could cause a CPU to fail orthos in one hour, then be perfectly fine for 60+hrs? Temps are 40c and below under load. Stock voltage & speed, used to be OCed (without any additional voltage)

HDD, spinrite: Passed
memtest: 3 loops passed
CPU: Fail?

More about : cpu fail orthos

a b à CPUs
July 17, 2009 2:37:57 AM

It prob wasn't seated right
a b à CPUs
July 17, 2009 2:43:11 AM

Upendra09 said:
It prob wasn't seated right

I didn't see any mention of the OP reseating the cpu, do they do that by themselves ?
Related resources
a b à CPUs
July 17, 2009 2:49:19 AM

delluser1 said:
I didn't see any mention of the OP reseating the cpu, do they do that by themselves ?


Yea, i know, but do you have any suggestions? :) 
a b à CPUs
July 17, 2009 2:15:43 PM

A few idea's , but nothing concrete, which is why I didn't respond to the OP's question.
I've had it happen before, run Orthos for 10 or 12 hours with no problem, start it up a couple days later and have it crash within 20 minutes.
Made no changes to the settings and ran it again the next day for 12 hours , no problem.
Some possibilities
Voltage surge or drop from the main ?
Flakey VRM ?
Ballistix ? after losing 5 sets in 3 different machines over the course of a couple months I won't use it anymore.
a b à CPUs
July 17, 2009 2:20:04 PM

what is orthos? never heard of it before
a b à CPUs
July 17, 2009 2:23:06 PM

It's a program used to test the stability of your overclock, Google it.
a b à CPUs
July 17, 2009 3:09:30 PM

ah i see, in that case maybe the software is doing something bad to the cpu
July 17, 2009 5:04:54 PM

I don't understand the recent great interest in subjecting pc to excessive and unnecessary stress.

Solution = don't run orthos. Perhaps I am missing something?

It seems a lot of peeps are actually buying pc's based on useless stressers like Prime, which is useless anyway. The driving force seems to be esoteric enlargement of one's epeen. :)  or the illusion thereof.

Perhaps there are also good reasons for such tests; but the general public's indulgence in this area is hardly required or useful. Perhaps this thread is aimed at engineers or IT sys admins - in which case, there might be usefulness to all this; if so I apologize for this diversion, and I am sure you see my point.
July 17, 2009 5:30:41 PM

sighQ2,

If a processor fails any stress program, it is guaranteed that you will get some errors (application stops responding, blue screen and restarts) at some time.

If you ignore the results of a stress test:

Firefox or any browser may crash once a week for example, and you will not know if it is firefox's problem or a hardware problem.

A game will crash after some hours of play. You will not know if it is a game-bug or a hardware problem. You may blame the game, you may blame the drivers, put actually most times it is a hardware failure and many people don't know it.

A great example of such a situation was when games with multi-thread/multi-core support started to appear, there were various posts-complains on forums of such games crashing all the time when their other games (single threaded most times) never did. Most people blamed the developers or the graphics drivers, but in reallity most of the time it was their hardware failing, as it was the first time their CPU+GPU where fully used and actually stressed enough. They thought they had a stable system when they did not.

Passing even extreme tests is not rare at all. I bet that 95% of all the (non-overclocked) PCs in the world are rock stable whatever the stress. But you never know if you are in the 5%, and in case of overclocked systems (even slight oc), it is a must. Anyone can overclock with instability. The rock-stable overclock, is the only overclock worth doing.
July 17, 2009 7:01:34 PM

I do not usually deliberately create trouble on my pc. If I have trouble, I can go looking for it, and perhaps these tests would be useful for tshoots. My way is to just use the pc for what I do anyway. If an oclox or other tweak helps and runs well, I am ok. But I don't see why I would want to break my pc just to prove it's ok. Which is just making a funny of it.

My point is more about people buying pc's and then obseesing over prime results, or many other useless benches like vantage /other crawp - just to prove that I am enough - that mine is as big as yours - that I am man enough - that my status is the community is now assured - that I am king of the hill, etc etc etc - it's all useless and sadly comical.

I understand how these things might be useful - I also am saying, in most of the situations where they are being used, that they are a waste of electricity, and nothing is gained. If it's working well; I am ok right there. I don't need to indulge in excessive "burn in". I have no use for going there. And beyond that; most benchmarketing is a cherry pick to illustrate superiority, which might not be the case in real world daily usefulness and functionality. Kinda like, it ain't broke; so don't fix it; and moreso, don't break it. But to each his own - most peeps doing this are just saying wowee, look at me and mine - and then go on and use it mysteriously without the benchmark running - this is not rocket science. People will continue with this regardless of what I think - that's ok, count me out.
a b à CPUs
July 17, 2009 7:16:08 PM

sighQ2 said:
Perhaps I am missing something?

It seems a lot of peeps are actually buying pc's based on useless stressers like Prime, which is useless anyway.

Apparently
I've never seen or heard of anyone purchasing a computer based on Prime.
It's not a benchmarking tool and is not useless to those wanting to make sure that thier overclock is stable.
a b à CPUs
July 17, 2009 7:26:53 PM

Upendra09 said:
what is orthos? never heard of it before



then do not give advice.

my computer can run perfectly fine with basic windows operations with an unstable overclock. its only when i run games or fold that i run into problems. orthos will definitely show you if you are stable or not - but RAM is another issue.
a b à CPUs
July 17, 2009 7:27:44 PM

GNR said:
sighQ2,

If a processor fails any stress program, it is guaranteed that you will get some errors (application stops responding, blue screen and restarts) at some time.

If you ignore the results of a stress test:

Firefox or any browser may crash once a week for example, and you will not know if it is firefox's problem or a hardware problem.

A game will crash after some hours of play. You will not know if it is a game-bug or a hardware problem. You may blame the game, you may blame the drivers, put actually most times it is a hardware failure and many people don't know it.

A great example of such a situation was when games with multi-thread/multi-core support started to appear, there were various posts-complains on forums of such games crashing all the time when their other games (single threaded most times) never did. Most people blamed the developers or the graphics drivers, but in reallity most of the time it was their hardware failing, as it was the first time their CPU+GPU where fully used and actually stressed enough. They thought they had a stable system when they did not.

Passing even extreme tests is not rare at all. I bet that 95% of all the (non-overclocked) PCs in the world are rock stable whatever the stress. But you never know if you are in the 5%, and in case of overclocked systems (even slight oc), it is a must. Anyone can overclock with instability. The rock-stable overclock, is the only overclock worth doing.


no exactly.

i can pass a stress test just fine - firefox will still crash. RAM testing is just as crucial as CPU testing.
a b à CPUs
July 17, 2009 7:37:06 PM

werxen said:
no exactly.

i can pass a stress test just fine - firefox will still crash. RAM testing is just as crucial as CPU testing.

I would consider RAM testing to be part of any good stress testing setup. I usually use Linpack to test RAM and CPU at the same time - that guarantees that your overclock is truly stable (though admittedly, Linpack stresses your CPU beyond any normal level).
a b à CPUs
July 17, 2009 7:49:25 PM

same here. though many people associate stress test with just cpu which is not true by any means. my overclock at 4.2 was stable on orthos but had multiple program crashes associated with ram.
July 17, 2009 11:09:18 PM

ok so back to my problem guys lol. Yes.. I have been getting increasing crashes in Firefox, IE, chrome, random corruption, trouble installing, BSOD... they kept getting more freuquent. It was like once a week. Assumed it was software issues. Then it got to be multiple times a day. So i reloaded windows, and still started to get firefox to crash.... Now all of the sudden on my second run of orthos its been rock solid... Anyways. I have to reload windows cuz I used the wrong installation disc (non legit) So if I am still getting crashing issues, I will address this topic again. Till then. You can keep quibbling over who should and shouldn't be using orthos and stressing their PCs :-)


BTW I am using ballistix ram... as one of the users posted. Could be the issue. It was the last upgrade I made to my system went from 2 gigs to 8 gigs.
July 18, 2009 12:00:44 AM

Crashing in normal software is not about heavy load or stress testing. A memtest might be useful for you however. But if you crash in firefox, I would expect total chaos in a stress test. I think you have some other setup issues, connections, drivers, compatibility, - like that.

If a 2nd run of a test runs better, it is because the first run caches the program, and it's already cached for the 2nd run - so an easier test. Actually this a form of cheating that certain review sites use to fudge results or winners or otherwise misrepresent - eg nanandotek or whatever it's called - I don't go there if at all possible.

I would, first thought, suggest video drivers - 2nd thought mobo ram specs re using 4 dimms or 8 gigs. Just an idea. Always check all connections. And check all cooling-related hardware to eliminate ovheat concerns. Also oclox settings if any, volts etc. Possibly an rma case. who knows. PSU? mobo?
a c 172 à CPUs
July 18, 2009 5:02:54 PM

If Orthos (dual core)/Prime95 (quad core) fails, the problem could be in the CPU, RAM, motherboard, or even power supply. It could be because of poor thermal management (overheating). They can simply fail, or they can crash and cause a reboot.

Some people do not stress test at all. Some stress test long enough to max out the CPU temps. Some test some arbitrary number of hours - 2, 4 , 8, or 12 hours.
A few of us test for 24 hours.

Orthos and Prime are not benchmarks. They are designed to load the CPU's to 100 per cent, where among other things, thermal problems will become very apparent.

"If a 2nd run of a test runs better, it is because the first run caches the program, and it's already cached for the 2nd run - so an easier test. Actually this a form of cheating that certain review sites use to fudge results or winners or otherwise misrepresent - eg nanandotek or whatever it's called - I don't go there if at all possible. "

A properly designed test program should flush the cache before it starts each time. But that is a separate issue from stress testing. Prime95 small fft test is designed to run from the cache. If a CPU has to go to main memory for something, it's not running at 100%.

Orthos and Prime95 will run fine without video drivers, depending on just the generic Windows drivers.

"And check all cooling-related hardware to eliminate ovheat concerns. Also oclox settings if any, volts etc. Possibly an rma case. who knows. PSU? mobo?"

Well, yes. And that's the whole point of stress testing - to work the system hard enough to bring out any of these problems.
!