1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

best way to stress cpu for stability

Discussion in 'Overclocking & Cooling' started by exodusprime1337, Mar 2, 2008.

  1. exodusprime1337

    exodusprime1337

    Joined:
    Nov 27, 2007
    Messages:
    2,188 (0.89/day)
    Thanks Received:
    342
    so i'm using orthos doing priority 10 running at 3.405 1.525v. on the small ffts-stress cpu, should up the lock a bit more and try some more stressing or should i go longer?? i'm hoping for about 3.45-3.5 or so max oc??
  2. Duxx Guest

    Not quite sure what you are asking... but for complete stability, i usually let it run over night and if i wake up in the morning and its still running, its stable. Between OC's i usually stress for 20-30 then when i get somewhere i'm satisfied, ill let run over night
  3. exodusprime1337

    exodusprime1337

    Joined:
    Nov 27, 2007
    Messages:
    2,188 (0.89/day)
    Thanks Received:
    342
    cool cool, yeah i've been thinking my 5000 can get a little further, current idles at 28c and loads at 40c with no ihs and a lapped ultra120 xtreme. i'm just bored so i figured i'd try and squeeze a few more mhz out of this thing, currently 30 minutes in at 3.416Ghz, should i move up another notch and keep going?? btw this is at 1.55v prolly not needed but this thing runns cool now that i got that ihs off?
  4. Duxx Guest

    Temps seem actually rather good with vcore at 1.55, but depending on how long you want to keep your 5000, id find a max, short of 1.6v and use it for benching only. You won't hardly notice a difference and their isn't a reason to keep it 24/7 with it that high.
  5. exodusprime1337

    exodusprime1337

    Joined:
    Nov 27, 2007
    Messages:
    2,188 (0.89/day)
    Thanks Received:
    342
    i'm aware of that, i dont' intend on going anywhere over 1.55 i want the highest oc i can get at 1.55. i upped it from 3.416 to 3.444Ghz and i'm starting for another 1/2 hour and well see what we get there
  6. Wile E

    Wile E Power User

    Joined:
    Oct 1, 2006
    Messages:
    24,324 (8.44/day)
    Thanks Received:
    3,778
    Orthos also isn't the last word in stability. I've had it error and stop, after running fine for 9 hrs. I've also had it where my machine was Orthos stable for 24hrs, but was still unstable with random app crashes. Orthos can give you a rough idea, but the only true test of stability is just using the machine as you normally would, and see if anything crashes.
  7. SirKeldon

    SirKeldon New Member

    Joined:
    Feb 7, 2008
    Messages:
    489 (0.20/day)
    Thanks Received:
    154
    Location:
    Barcelona, Catalonia (Spain)
    I'm agree with Wile E, i think all the programs are meant to be buggy, that's not an offense to the programmers, cause i'll be throwing rocks over my shoulder too, but errors happens and even at stock speeds or with rough and stable overclocking, some programs won't run or will crash "magicly", informatics are not a exact science like maths ... it's more like physics, freqs are changing every second and also you can't control the memory at all, the allocation it's an OS related affair.

    I've found best ones for AMD's right now are Prime v25.6 (though it's a beta yet, has a few bugs solved, you can find it here) and the AMD overdrive, for me it's the best one, we'll be agree it's not the best "stresser" as it adjusts itself the preference of the CPU to run ... but as a background stability test it's the best and also is good for a high-end stability test with no other apps running. Yesterday night i gave a try to Prime95 v25.5 (stable version) ... CPU supported the Blend Test for 11 hours at 3360 1,392V ... and today, i was installing NFS ProStreet to test overclock and boom, crashed after 36 hours powered on (with a few reboots) ... i thought "wtf? prime it's not the best one?".

    So as Wile E says, best chance is to use the PC as you will like to use if it was stable ... most of overclocks likes to fail or crashing the first week, that's the "best stability test" i will do, and that's what i'm doing: running games, encoding videos, uncompressing big files, treating large images with huge filters on photoshop, testing audio encoding and audio live effects, hardware's related stuff ... and more ... so keep trying my friend :toast:

    ps: NFS Pro-Street is a good game to stress the CPU while i'm doing another things? I liked the NFS series and i downloaded the last one ... omg, i'm outta the game scene for almost two years :laugh:
    Last edited: Mar 2, 2008
  8. exodusprime1337

    exodusprime1337

    Joined:
    Nov 27, 2007
    Messages:
    2,188 (0.89/day)
    Thanks Received:
    342
    someone please help me with my oc.. so i went the last hour or so with 3.46 1.525v blend test with mem turned down. well i figured i'd bring my mem back into the picture and bam i got errors. so here is the deal

    3.36Ghz 1.45v: 240x14 mem ddr960 4,4,4,13 2t-12hours orthos stable-current oc

    3.47Ghz 1.50v: 248x14 mem ddr992 5,5,5,15 2t -2 hours orthos stable

    here is the problem, in order for me to hit 3.47 i have to loosen up the timings by a large amount and in all honesty my bench scores practically go down. what do i do now?? in order for me to hit a higher tier oc i need to hit the 1100's for mem?? should i get some 1066 crucial and hope it oc's to 1100
  9. Wile E

    Wile E Power User

    Joined:
    Oct 1, 2006
    Messages:
    24,324 (8.44/day)
    Thanks Received:
    3,778
    Try 3.4 even at 1.5V, and see if your ram will run at 970 4-4-4-12. If the ram won't, raise it's voltage a notch, and try again. Don't run the ram above 2.3V for 24/7 use tho.

    Your chip clocks almost exactly like my 6400+ does. Mine went to 3470 1.5V, but not stable. Ran perfectly at 3.4 1.5V 24/7. Not a single crash on those settings for months.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page