I am using Linpack testing in OCCT's package to stress my CPU. I am doing a 10 minute test, but is this long enough and is Linpack (based on Intel Burn Test) a good means of judging stability?
Is OCCT Linpack testing a good measure of stability?
98uk
1 - 18 of 18 Posts
Joined
·
1,250 Posts
I too use the linpack test in OCCT along with the OCCT stress test....I Usually run the Linpack test for 1 hour straight and so far have not had any issue's from an overclock that I did not cause.
Joined
·
1,005 Posts
A lot of people say that IBU (Linpack) puts unnecessary stress on your components. Under normal use, your computer would never be exposed to those conditions. This means that if you pass Linpack, your computer is for sure good, but htat if you fail it, your computer might still be good enough for normal use.
A lot of people still prefer running prime95 for a longer period to more accurately guage stability.
A lot of people still prefer running prime95 for a longer period to more accurately guage stability.
Joined
·
719 Posts
Nothing is a better measure for stability than your daily usage of your system. Do the things you actually do with it. Play games, run some apps, encode some movies, do your regular work bla bla bla and if you are having no problems or performance issues with it than it means it is stable. There is no such application in the real world that will stress your cpu that much as those stress tests for a that long period of time.
Joined
·
1,005 Posts
Well the problem with just doing daily tasks to test stability is that it might not stres the components enough so that it will cause system crashes etc, but may cause corrupted files etc, which would be hard to notice on a normal use basis.
Joined
·
719 Posts
Quote:
If it would cause system crashes then you would know something is wrong and that your system is not stable and would go back to tweaking. You can also have system crashes while prime95, linx, IBT or OCCT is running. I have never heard of a bad oc that cause corrupted files. If it is not stable your daily usage would give the signals for its unstability.
I have ran all these tests day and night then strated to find it pretty funny when I saw couple of chips even failing these tests at stock settings although causing no problems with daily usage for years.
Originally Posted by stevenma188  Well the problem with just doing daily tasks to test stability is that it might not stres the components enough so that it will cause system crashes etc, but may cause corrupted files etc, which would be hard to notice on a normal use basis. |
I have ran all these tests day and night then strated to find it pretty funny when I saw couple of chips even failing these tests at stock settings although causing no problems with daily usage for years.
Quote:
It all depends on how quickly you want to find out if there is instability in the system. If it needs playing games, encoding a video while watching a movie, editing a photo or video and writing a doc and browsing the net, it will take me a couple of days to figure if that BLCK jump from 160 to 162 is stable or not. With LinX, typically it dies within 1-2 hours if the system is not stable. Economies of scale!
BTW, if any of these tests fail: there is a problem and the system will die on you randomly even if you have successfully played games for 3 days straight on without a crash. It will die at a very in-opprtune moment (like that special scene in the movie or just before you go to save that doc you were working on). The thing about computers is that they behave in a deterministic fashion: you pass same data, it will give you same results. If the results are different once in a million times, your computer is a fail, and you need to go back and figure which component (cpu, memory, PSU, NB etc.) has failed you.
So, make sure after the final OC is settled on, you pass 100 passes of LinX (which btw is the best cpu and memory tester) with max memory setting on a 64-bit OS.
Originally Posted by Prong If it would cause system crashes then you would know something is wrong and that your system is not stable and would go back to tweaking. You can also have system crashes while prime95, linx, IBT or OCCT is running. I have never heard of a bad oc that cause corrupted files. If it is not stable your daily usage would give the signals for its unstability. I have ran all these tests day and night then strated to find it pretty funny when I saw couple of chips even failing these tests at stock settings although causing no problems with daily usage for years. |
BTW, if any of these tests fail: there is a problem and the system will die on you randomly even if you have successfully played games for 3 days straight on without a crash. It will die at a very in-opprtune moment (like that special scene in the movie or just before you go to save that doc you were working on). The thing about computers is that they behave in a deterministic fashion: you pass same data, it will give you same results. If the results are different once in a million times, your computer is a fail, and you need to go back and figure which component (cpu, memory, PSU, NB etc.) has failed you.
So, make sure after the final OC is settled on, you pass 100 passes of LinX (which btw is the best cpu and memory tester) with max memory setting on a 64-bit OS.
Joined
·
1,005 Posts
Couldnt have said it better, +REP
Joined
·
719 Posts
Quote:
100 runs of Linx ? Wow. I remember running Linx 20-30 runs but 100, I don't think so. My cpu would never forgive me for torturing it in such a way. It looks like a an another way to fasten its degradation process. Linx seriously gets your chip hot and I don't think running linx for hours at 60-70C's is a good idea. 20 runs is more than enough with Linx IMO.
Originally Posted by devsk It all depends on how quickly you want to find out if there is instability in the system. If it needs playing games, encoding a video while watching a movie, editing a photo or video and writing a doc and browsing the net, it will take me a couple of days to figure if that BLCK jump from 160 to 162 is stable or not. With LinX, typically it dies within 1-2 hours if the system is not stable. Economies of scale! BTW, if any of these tests fail: there is a problem and the system will die on you randomly even if you have successfully played games for 3 days straight on without a crash. It will die at a very in-opprtune moment (like that special scene in the movie or just before you go to save that doc you were working on). The thing about computers is that they behave in a deterministic fashion: you pass same data, it will give you same results. If the results are different once in a million times, your computer is a fail, and you need to go back and figure which component (cpu, memory, PSU, NB etc.) has failed you. So, make sure after the final OC is settled on, you pass 100 passes of LinX (which btw is the best cpu and memory tester) with max memory setting on a 64-bit OS. |
Its a machine. It doesn't feel a thing. You feed it volts and it works.
Anything less than 4 hours (translates into about 50 runs with 64-bit on 6GB memory) of LinX is not stable in my eyes. Unless and until you have $50 mobo (if that's even possible with i7) or really bad cooling, you should not worry about stressing the system.
Anything less than 4 hours (translates into about 50 runs with 64-bit on 6GB memory) of LinX is not stable in my eyes. Unless and until you have $50 mobo (if that's even possible with i7) or really bad cooling, you should not worry about stressing the system.
I ran OCCT v3.1 -
CPU:OCCT, success (medium & large data set)!
CPU:LINPACK, failed, error detected.
So I give IBT a go with standard 1024MB, and everything is fine.
Then I go for the very high which uses 4096MB ram, and failed.
Should I suspect my ram? But I tested it with memtestx86 and memtest, both return no errors.
CPU:OCCT, success (medium & large data set)!
CPU:LINPACK, failed, error detected.
So I give IBT a go with standard 1024MB, and everything is fine.
Then I go for the very high which uses 4096MB ram, and failed.
Should I suspect my ram? But I tested it with memtestx86 and memtest, both return no errors.
Joined
·
843 Posts
Quote:
I have to agree with Devsk. Stability means your computer can handle the absolute toughest assignment you can throw at it. Anything less is just conditional stability. I'm not okay with conditional stability. Some users are. It's personal preference, I suppose.
Originally Posted by Prong 100 runs of Linx ? Wow. I remember running Linx 20-30 runs but 100, I don't think so. My cpu would never forgive me for torturing it in such a way. It looks like a an another way to fasten its degradation process. Linx seriously gets your chip hot and I don't think running linx for hours at 60-70C's is a good idea. 20 runs is more than enough with Linx IMO. |
OCCT Large Data @ High Priority is good way to test stability. It's equal load as Encoding video while playing a 3D game & running Fraps. So I let it run for 2 hours and if it past, it's rock solid
![Image]()
LinPack is as I've always said, is a bit misleading. Same as Orthos. These programs run 100% COSNTANT slam on memory & cpu as if it had no chance to relax. Even while you run a 3D game, play winamp & FRAPPING it's still gona be 95% - 100% randomly per each .5 sec & not like ORTHOS or LINPACK suggests running 100% load at 100% timing..
LinPack is as I've always said, is a bit misleading. Same as Orthos. These programs run 100% COSNTANT slam on memory & cpu as if it had no chance to relax. Even while you run a 3D game, play winamp & FRAPPING it's still gona be 95% - 100% randomly per each .5 sec & not like ORTHOS or LINPACK suggests running 100% load at 100% timing..
Quote:
100? I run 300+ And then do 18+ hours of Prime Blend.
![Image]()
I am paranoid about stability.
Originally Posted by Prong 100 runs of Linx ? Wow. I remember running Linx 20-30 runs but 100, I don't think so. My cpu would never forgive me for torturing it in such a way. It looks like a an another way to fasten its degradation process. Linx seriously gets your chip hot and I don't think running linx for hours at 60-70C's is a good idea. 20 runs is more than enough with Linx IMO. |
I am paranoid about stability.
Joined
·
1 Posts
Obviously.
1 - 18 of 18 Posts
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
