• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** Official Opteron 170 Ocing Thread ***

Clairvoyant said:
So would you say 100 c + is ok, if i closk to 2.8 my Chipset goes over 100 c
Click to expand...

Fair play, now that’s what I call toasty! I think you will have issues with the usable life of your board if your chipset runs constantly at that temp, silicon can only take so much heat for so long when its cut that thin.
Pulse width modulator ic's can take + of 100c yeah but is he talkin about the pwm's?
 
Scoobie Dave said:
Really you should use large FFT's rather than small FFT's, large will strain your memory as well as the cpu. If it can pass 12+ hours of that then you can class it as 'prime stable'. ;)
Click to expand...

Surely you want blend for memory and then why test memory with prime? I always stick memory on a divider and then clock cpu to the max and test with small fft's. Then get memory up to its max at the timings you want by using memtest. Then combine both cpu and memory on or as close (but below) the max settings. Ultimate test after that has got to surely be 3dmark loops as that tests the whole system in game mode.

I know this is just my method of testing and everyone has their own method.
 
ted34 said:
Surely you want blend for memory and then why test memory with prime? I always stick memory on a divider and then clock cpu to the max and test with small fft's. Then get memory up to its max at the timings you want by using memtest. Then combine both cpu and memory on or as close (but below) the max settings. Ultimate test after that has got to surely be 3dmark loops as that tests the whole system in game mode.

I know this is just my method of testing and everyone has their own method.
Click to expand...

I agree,

Large FFT's used to be the thing, then everyone was sayin you need to do small FFT's to push the CPU to its max. I realised 4 hours isnt great but i got bored and had to play guildwars factions again ;) I was just happy to get over 2.8Ghz for once, 10 processors on.
When I find my max clock ill do a 12 hour then post again. Was hoping for 3Gig but I dont think this wil quite make it.
 
Devious said:
I agree,

Large FFT's used to be the thing, then everyone was sayin you need to do small FFT's to push the CPU to its max. I realised 4 hours isnt great but i got bored and had to play guildwars factions again ;) I was just happy to get over 2.8Ghz for once, 10 processors on.
When I find my max clock ill do a 12 hour then post again. Was hoping for 3Gig but I dont think this wil quite make it.
Click to expand...

Be happy with 2.8, I doubt you would really notice the difference running 3gig anyway. . .
My chip has plenty more in it, but im happy running it @2.8 with only 1.36v. ;)
 
ted34 said:
Surely you want blend for memory and then why test memory with prime? I always stick memory on a divider and then clock cpu to the max and test with small fft's. Then get memory up to its max at the timings you want by using memtest. Then combine both cpu and memory on or as close (but below) the max settings. Ultimate test after that has got to surely be 3dmark loops as that tests the whole system in game mode.

I know this is just my method of testing and everyone has their own method.
Click to expand...

I thought blend was a combinatioin of cpu&memory tests?

Anyway like you said everyone has there own method. I tend to do what the majority do :D .
 
like ted says if you have your ram on a divider why stress test it, large ffts and blend seem to give the cpu an easy time. but is good for lazy people who like to test cpu and ram at the same time. from my experience even though small fft tends to generate less heat it seems to generally push the cpu further. 1 hour of large ffts usually equals 1 min of small ffts. that how I've found it with my systems anyway.
 
Last edited:
Hesky82 said:
like ted says if you have your ram on a divider why stress test it, large ffts and blend seem to give the cpu an easy time. but is good for lazy people who like to test cpu and ram at the same time. from my experience even though small fft tends to generate less heat it seems to generally push the cpu further. 1 hour of large ffts usually equals 1 min of small ffts. that how I've found it with my systems anyway.
Click to expand...
Because you aren't only stressing the RAM - it's more of a memory controller test than RAM. You can trust me on this one - 16h prime stable of small FTT's fell over after 1 minute of large FTT's.
 
Scoobie Dave said:
Cheers for backing me up dude ;)

Like you said, I find large FFT's for more stressful.
Click to expand...
Most definitely. I've had 3 CPU's pass 15h+ on small only to crash out of UT2004 after about 30mins with a 'General Protection Fault' - I then ran large FTT's which failed in 2 minutes. You have to run Large FTT's to ensure stability. Small FTT's will tell you if the CPU itself is fine at those speeds (as it fills up the cache) but it will not tell you the state of the memory controller - i.e. the thing that holds most people back with A64's.
 
smids said:
Because you aren't only stressing the RAM - it's more of a memory controller test than RAM. You can trust me on this one - 16h prime stable of small FTT's fell over after 1 minute of large FTT's.
Click to expand...

depends what hardware/area is the weak link in your system, I run all torture tests and run the one that fails the soonest and go by that, found that small fft fails way before large ffts on my current system, you cannot say one torture test is more stressful than the other because they test different things, and pick out instability in different areas.
 
28hrPrimeStableOpty2008test.jpg


28 hour prime session. Stable enough I think! :D
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom