View Full Version : OCCT or Prime95
Which do you believe is better for the ultimate CPU stability testing, OCCT or Prime ?
Smartidiot89
03-20-2009, 11:05 PM
Personally I use LinX, but OCCT is lightyears ahead of Prime95.
Personally I use LinX, but OCCT is lightyears ahead of Prime95.
Care to elaborate as to why it is "light years ahead" in terms of CPU stability testing.
newls1
03-21-2009, 07:58 AM
If I use OCCT for the 30Min CPU stress test, I'm completly stable. If I use P95, I fail and lock up with in 5 minutes. So I ask "SMARTIDIOT", why is OCCT "LIGHT YEARS" ahead of P95?
SoulsCollective
03-21-2009, 08:11 AM
Neither - some variation on Linpack. OCCT is reasonably good at whole-system-load, but can fail to pick up on some errors that will show in P95, which is generally more stressful in the small FFTs mode. However, the OCCT functionality and monitoring plugins are worlds ahead of P95. In my experience, though, the ultimate test is a good, long Linpack run - drives temps higher, and will show up CPU instability well before either OCCT or P95.
Linpak is really really extreme, ive been stable at 3.2Ghz at 1.335v for a year but it fail in linpak, i need 1.355 for it
Leeghoofd
03-25-2009, 07:06 AM
There's no ultimate testing tool , do a combination of all if required (hyperpi, OCCT, Prime, IBT, 3DMARKS, WPRIME...)
Best stability test is daily usage !
LOUISSSSS
04-20-2009, 08:01 PM
There's no ultimate testing tool , do a combination of all if required (hyperpi, OCCT, Prime, IBT, 3DMARKS, WPRIME...)
Best stability test is daily usage !
no its not. that is how you get a corrupted OS; by using unstable hardware
Leeghoofd
04-21-2009, 08:54 AM
First it has to pass some stress tests of preference and then run the rig and see if it craps out somewhere.
I usually do some Test 3,4 and 7 of Memtest, before I even boot into Windows at high FSB settings (to avoid corrupt Osses). From there on I run some Linx ( 10 passes ) Though for me it doesn't need to pass that one as no application will stress my CPU as hard as that utility does. If it caps out on one iteneration I'm pretty sure te Vcore is okayish. Then I run some Blend test (8 hours), Hyperpi 32Mb ( on all cores ) and finally some 3dmarks. My rigs fold a lot and if I don't get errors there then I could consider it stable for me. But I never ran 50 passes Linpack, neither 24 hours prime or co and yet the rigs seem to hold up pretty well.
You seem to miss my point : what I said was that there is no conclusive testing utility. Some tests are better for the CPU , some are better for NB, others for ram testing, some 3D... if it passes test 1,2 and 3, but not test 4 : Is it unstable then for daily usage ? dunno you tell me...
It seems these discussions seem to be endless with the linpack fans... I'm not telling one test program or the other is better. I'm just telling that they all have their own advantages and disadvantages...