On Using the Prime95 Torture Test with an AMD Phenom

[rabrittain]

Thanks for the info, VirtualLarry!!

I have been considering a different case and more fans. That is an interesting little mod that I haven't seen before. I like the website, too.

Have you seen my memory?

http://www.ocztechnology.com/products/memory/ocz_ddr2_pc2_8500_reaper_hpc_4gb_edition

I'm not sure if the ram cooler would fit over the heat pipe.

I'm concerned about summer, though. I live near Tacoma, WA, and we really don't get that much hot weather here. We may have 2 weeks of weather with highs in the 90's in July or August. I have no air conditioning. So, with the computers running, my apartment turns in to an oven. It's nice in the winter. I hardly ever have to turn on the heat. In the summer though .... (Again, it'll be gone almost before I realize it's here.) It makes it hard to sleep, though. I like it cold where I sleep. During that time, I really have to watch the computers to make sure they don't overheat. Rudy Toody says that I should consider water cooling. Maybe he's right.

 

[Assimilator1]

Seeing as your RAM has heatpipes on it I can't see that it'll need anymore cooling (unless you put insane voltages through them, which you're not).
So are you saying you've pinpointed 1 RAM module as faulty?

I can't remember what the temp limits are for Phen II but it sounds like it's running rather hot. Maybe this is the problem?

Yeah... but this is OCZ we're talking about...

Lol, now ya got me worried as both my rigs have it!
I thought OCZ was a decent make, what's wrong with it? (although I have had problems with the OCZ DDR2 800 CL4 stuff, 1st pair - one stick went bad on me, the 2nd pair no longer runs at CL4 & now has to run at CL5 , although I haven't tried higher voltages.)

Btw your RAM coolers link was dead for me.

 

[Rudy Toody]

Lol, now ya got me worried as both my rigs have it!
I thought OCZ was a decent make, what's wrong with it.

Four years ago I read in one review that OCZ was making a concerted effort to overcome their bad rep.

As far as I can tell, they have done it!

I have used OCZ memory for 4 years without any problems and I have used OCZ Vertex SSDs for the last year without problems.

Actually, one SSD's format got clobbered, but the OCZ techies helped me get it back. It has run solid for almost a year!

 

[rabrittain]

Seeing as your RAM has heatpipes on it I can't see that it'll need anymore cooling (unless you put insane voltages through them, which you're not).
So are you saying you've pinpointed 1 RAM module as faulty?

Yes! There was poor air circulation inside the case, especially around the RAM. When I tested 1 dimm at a time, one gave trouble, and the other did not. I have marked them, and I know which one gave trouble. However, after I made the changes, there was no trouble. For the time being, I consider the problem solved. When I get some real life stuff out of the way I'm going to crank the cpu back up to at least 3.3 GHz and see how the system does.

If everything goes as planned, I'll be employed before the end of September; then, I can buy a better case with better cooling options. After that, I may consider running the cables behind the motherboard and a better cpu cooler. It would be nice to try to see if I can boost the system to a stable 4.0 GHz.

 

[Assimilator1]

That would be a good o/c!

How long a test have you done with the 2 RAM sticks at 1066 now?
Hopefully you've got it sussed then :thumbsup:

 

[rabrittain]

I ran the Prime95 torture tests through two sets of tests, FFT length = 1024k and FFT = 80K. It did ok, so I put it back online with Rosetta. I guess that the real test will be whether of not it'll run Primegrid. I should find out before too long.

 

[Ken g6]

Yeah, the July race will probably flush out any problems; but we don't have another nasty race like the last one until the end of September.

 

[Assimilator1]

Well a 24hr run of P95 is the real test, although I use OCCT nowadays as it picks up errors more quickly.

 

[Mr. Pedantic]

Well a 24hr run of P95 is the real test, although I use OCCT nowadays as it picks up errors more quickly.

It is not. I've had overclocks that were 30+ hours P95 stable that gave EUE errors with the SMP client within minutes. And that's not even -bigadv.

 

[Assimilator1]

Yea now you mention it, I remember I have too! :$ (although not in minutes), that's why I switched to OCCT in the 1st place, lol.
No EUEs after testing with OCCT....... yet , have you tried OCCT?

Although right now I'm skipping the stabilty test (less DC downtime) with the new speed/vcore settings my Q6600 has & I'm just watching out for EUEs!
So far 24hrs & no EUEs, I've not decided yet if this is a good idea to do it this way though .....
(I did copy the entire F@H dir incase of an EUE in the then current WU).

 

[rabrittain]

I have succesfully overclocked the computer - a little bit. It is stable at 3.2 GHz. The cpu voltage is normal (1.35 v).

I may have to check out occt.

 

[Assimilator1]

Cool, what temps does it run at?

 

[rabrittain]

Sorry Assimilator1 - I'm not ignoring you - I've been busy with other things.

My notes are at my apartment, and I'm not there right now. Itry to give you a few numbers from memory.

I first tried to overclock the machine to 3.4 GHz. The cpu voltage was set to 1.40 volts. It booted just fine, but the Prime95 torture test dropped out after about 50 minutes. As best as I can remember - the core temperature was around 70 degrees C. I dropped the cpu voltage to 1.35 and tried again at 3.3 GHz. The torture test dropped a core after about an hour. The core temperature was down below 60 deg. C. I tried again at 3.2 GHz, and the torture test ran for over 2 hours. The core temperature was below 60 degrees C. I called it good, and that is where it has stayed. The machine has been crunching Rosetta and Simap with no problems.

 

[Assimilator1]

No worries, I've been busy too .

Sounds good, what's their safe limits?