Sign in with
Sign up | Sign in
Your question

Prime 95

Last response: in Overclocking
Share
April 10, 2010 11:52:42 PM

I have been working on OCing my sig rig for a while now I have hit another snag. My cpu was OCed to 272x14 @ 1.5v, with my HT set at x4 and my NB set at x4 and my memory set at 1:1(so 272) @ 5-5-5-15-20. The memory was mem test stable.

Well I ran prime95 small FFT for 12 hours and never got an error of BSOD or anything. I have done this a few of times, but when I run the blend test I get an error after about 30 minutes or so.

Is my cpu causing the error is could t be something else?

More about : prime

a b à CPUs
a b K Overclocking
April 11, 2010 12:53:40 AM

Quote:
Blend test focuses on the RAM, but it is far worse than Memtest86+.

Small FFT focuses on the CPU.

Large FFT focuses on the overall stability


If you pass Small FFT but fail Large FFT, the CPU could still be the cause of instability. FYI, I had experience in failing Large
FFTs in an hour several times after passing 24hours Small FFT and it was due to insufficient CPU core voltage. Hence, Large
FFT test is the trustworthiest test in Prime95 IMO.

p.s. LinX is also a very good stability test program.
m
0
l
April 13, 2010 10:40:43 PM

272x14 was 24hour small FFT stable but it is unstable on large FFTs. My understanding of these 2 tests was that small was able to fit all the calculations into the cpu cache, therefore stressing pretty much only the cpu. Large was too big and so memory was used to help store data for the calculations. So it appears that the error happens somewhere in communication between the cpu and mem. Do you think turning up the voltage on the HT or NB would help stabilize this or is it still just a cpu issue?
m
0
l
Related resources
a b à CPUs
a b K Overclocking
April 13, 2010 11:01:29 PM

BradyT88 said:
272x14 was 24hour small FFT stable but it is unstable on large FFTs. My understanding of these 2 tests was that small was able to fit all the calculations into the cpu cache, therefore stressing pretty much only the cpu. Large was too big and so memory was used to help store data for the calculations. So it appears that the error happens somewhere in communication between the cpu and mem. Do you think turning up the voltage on the HT or NB would help stabilize this or is it still just a cpu issue?


Quote:
If you pass Small FFT but fail Large FFT, the CPU could still be the cause of instability.

Hence, the instability could be due to CPU, HT or NB.

What is the maximum CPU core voltage on Phenom II? Try increasing it another one or two notch first if it is still within the spec.

*Tip*: The best way to find out which voltage is causing the instability is setting the voltage that might be too low according to your 6th sense to its maximum value with ALL other voltages staying unchanged and see if the system is stable after this. If it is stable, then it's the voltage you need to tweak. If it is not, try tweaking another voltage in the same way.
m
0
l
April 13, 2010 11:05:23 PM

The max safe voltage that AMD put out was 1.5v. I have mine set there and with my LoadLine Calibration cranked I hit 1.5-1.6v. I can't go any higher because temps start to break that 62C barrier at room temperature. HT and NB voltages are still at their stock and I have no idea what the maxes are. NB is currently 1.1v and HT is 1.2v.
m
0
l
a b à CPUs
a b K Overclocking
April 13, 2010 11:06:19 PM

BTW, you should paid a bit more for a PII-955 which comes with unlocked multiplier.

OCing is much easier with unlocked multiplier, because all you need to worry about would be the CPU core voltage.
m
0
l
April 13, 2010 11:09:39 PM

I know I wish I would have, but oh well. It hasn't been so bad so far and I've learned a hell of a lot along the way. This is pretty much a little system to get me by until September when I head off to Iraq for 12 months. Once i get back I will buy a whole new extreme rig, with the hopefully newly released architectures from AMD and Intel:) 
m
0
l
a b à CPUs
a b K Overclocking
April 13, 2010 11:11:52 PM

Increase your HT and NB voltages then. If it still doesn't work, it means that your chip is no good at that frequency and need to be lowered.
m
0
l
a b à CPUs
a b } Memory
a b K Overclocking
April 13, 2010 11:31:33 PM

Quote:
I have done this a few of times, but when I run the blend test I get an error after about 30 minutes or so.

Blend Test is RAM bound. My guess is you need to increase RAM voltage. Memtest86+ DOSEN'T really stress the RAM.

Also note: It could be a software,driver,etc conflict.
m
0
l
April 14, 2010 1:41:07 AM

I hope it isn't a driver/software issue and I doubt it is a mem voltage issue. The ram is ddr2 1200 5-5-5-15 and in cpuz it has an epp that is 533 5-5-5-15-43-2T @ 1.8v and a jedec that is 400 5-5-5-15-20-2T @ 1.8v. For these test I have the memory running at 272 (that is already doubled) 6-6-6-18-26-2T @ 2.1v. For OCing I managed to get them up to 1088 5-5-5-5-16-2T @ 2.1v.
m
0
l
April 14, 2010 4:32:17 PM

270x14 ran stable for 7 hours and so I cranked everything back up and got an error within seconds. So I left the CPU, HT and NB up and down-clocked the mem. I let that go for about an hour and it was stable. so down-clocked the CPU, HT and NB and turned up the mem. It got an error in about 10 minutes but it is memtest86 stable. I turned the NB voltage up from 1.1v to 1.45v (that was where the voltage was still green) and I am about to try that. Think that could fix the issue?

EDIT: Well that got an error in 7 minutes. I turned everything back up and I am just going to let memtest run for a few hours and see if it comes up with anything. I'm stuck at this point:( 
m
0
l
!