What Orthos CPU Stability Test do you run?

[Burn]

Well? Large FFT's, Small FFT's, Blend or Gromacs? I am trying to get a consensus on what everyone uses on the forums to test for CPU stability.

[Rattle]

blend always overnight at least

[shogo_ca]

for cpu stability only, Large FTT priority 9 for 8hour. For system stability... Blend for at least 8hour

[xlink]

All of them for 24 hours @ 3-5% over your normal 24/7 without voltage increases
then do OCCT and memtest likewise


I'm a stability freak...

[Ramaistro]

I have only used large fft's in my last overclocks and months later it is still stable. So i gues it is hard enough for my system to stress it with.

[citat3962]

I run small FFT to spot check for stability as I'm feeling something out but I tend to do overnight runs of blend to check stability on final settings.

[PanteraGSTK]

I use small fft's for just cpu stability, and blend for system stability.

[Millyons]

like i just posted in some other thread for c2d i only use blend .... small and large fft can be run for hours even stoped at 20+ h and blend would only last not even 10 min, and only vcore would solve it...first i was thinking mobo/ram voltage or timming but none of it helped only vcore got it stable

this is on 4 e6600, x6800, qx6700 and P5Bdlx and also my friends P5Wdh and his 6600

[NickS]

Small FFT's FTW

[oshox]

Blend should be used to test system stability If you get an error, then use Small FFT's to test if it's CPU or RAM related.

[justin_c]

yes. goldmemory/memtest for 3 hrs, then small fft 24hrs, then a few days with f@h in the background.

[ziddey]

I've had small and blend pass for 24+ hours (screenshots http://ijib.com/~ziddey/rig/ ) but running large, it'll usually crash the system after 14 or so.

[RubberDuck]

I have always used the custom option then set the program to use 1600 for the memory and let it run 7 -8hrs. Then run 2 instances with folding 24/7.

[MikalCarbine]

small ffts for cpu (8-128 IIRC) and anything higher stresses memory

I'll run small ffts if I am testing just my cpu clock, or blend for memory/system

[safan80]

use blend to test the whole system, Large the RAM, and small the cpu. rounding errors are the CPU and illegal SUMout is the memory.

[ziddey]

use blend to test the whole system, Large the RAM, and small the cpu. rounding errors are the CPU and illegal SUMout is the memory.

large to test cpu, including all cache up to 1mb per core, and up to 8mb of ram. no more.

[AntiNazi]

blend for 24hours.

[ziddey]

blend for 24hours.

puree for 36 hours

[SLi_dog]

I use Small FFTs to spot check for CPU stability and also run it for 10minutes to get an idea of the max CPU temps the system will experience. I then run Blend for at least 8hours. 24 hours would be perfect but I don't have the ambients to run for that long

I haven't had a system crash yet that had passed an 8 hour Blend test.

[trans am]

I make sure my memory is stable 1st. Then I crank the cpu until small fft errors out. then adjust vcore more. run small fft again and check if it can hold out longer. I do this until I get piss drunk and yell at my computer screen. Then if I'm out of beer and tired I'll go to bed and run Small fft while I'm sleeping. If I wake up to that annoying beeping sound I shut the computer off and deal with it in the morning. Or if I'm really tired I sleep through the beeps and wake up and my computer is frozen and temps are in the high 60s from all that vcore from the previous night. I'll use blend to test mem stability in windows after memtest86 passes a few times. I hope this was helpful.

[theteamaqua]

large

large stresses RAM, think of it this way

if ur CPU is unstable do u think that it can pass data correctly to RAM? thats why i mainly use large

blend doesnt stress RAM much, small stresses cpu the most , but doesnt stress RAM much

[Millyons]

large

large stresses RAM, think of it this way

if ur CPU is unstable do u think that it can pass data correctly to RAM? thats why i mainly use large

blend doesnt stress RAM much, small stresses cpu the most , but doesnt stress RAM much

only blend stresses ram, large doesnt either....most of the fft sizes can fit in modern cpu's cache

[lowfat]

24 hours of small, 8 hours of memtest, then 8 hours of blend

i like it really stable

[lowfat]

large

large stresses RAM, think of it this way

if ur CPU is unstable do u think that it can pass data correctly to RAM? thats why i mainly use large

blend doesnt stress RAM much, small stresses cpu the most , but doesnt stress RAM much

Blend uses a full 2GB, didn't know that wasn't much ram

[buffbiff21]

puree for 36 hours

I like to get it really stable as well as some others up there ^

For my current system, (which is really only my second system) The first thing I did was run Memtest for 24 hrs, my 1066 Dominators passed @ 1250 5-6-6-18 2.4 vdimm @ 800, I had them at 3-3-3-5 2.5 vdimm

FOr CPU, I use small FFTs, two instances. (although for stability "probing" I give it a quick Pi run). Then I get into teh Orthos. 5 hrs is enough for me. but I have passed 24 hrs at 3.6, and 5 hrs at 3.8 on my e6600.