• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Prime95 errors, what causes them?

S

Spewn

2[H]4U
Joined
Jan 5, 2001
Messages
3,459
My barton 2500+ can sit at full load at 2200mhz@1.725v forever and ever, no errors, at 11x200. If I try to go to 210, prime95 gives me a rounding error after a bit. CPU temps at full load at 2200mhz never go above 38'C, and at 2300 it never went above 40. It just seems kinda crappy that my cpu isn't really running hot at all but I can't seem to get anything more out of it :/ I went all the way up to 1.85v and still got the error.
 
Need some more info. What type of RAM and the timings? It could be that your RAM doesn't like to go that high if you have PC3200.
 
Ram timings at 200mhz fsb(and the ram is rated at pc3200) 8-4-4-3

Same timings for 210. If it is the ram, I would assume getting better ram/higher rated ram would let me overclock higher? It just seems like the chip *should* be able to go much higher.
 
Heat doesnt cause it, the CPU returning the wrong computation causes it. This happens from purely a too high OC. Not every chip is guaranteed to OC to a certan level, looks like you got the shortend of the stick.
 
If you have a locked multiplier on that Barton, get better memory. If not, pump up your multiplier and keep the mem at 200. Either way, eliminate the memory as the culprit first.
 
Unfortunately, it's multiplier locked :/ I'll definately consider getting better memory though.

Edit: I know heat doesn't cause prime95 errors, but increasing the voltage by .125 didn't make the errors go away, and heat wasn't even a problem at that point, that's what makes me think the chip has more in it than that.

From stress.txt(from prime95):
"The exact cause of a hardware problem can be very hard to find.

If you are not overclocking, the most likely cause is an Overheating CPU or memory DIMMs that are not quite up to spec. Another possibility is you might need a better power supply. Try running MotherBoard monitor and browse the forums above to see if your CPU is running too hot. If so, make sure the heat sink is properly attached, fans are operational, and air flow inside the case is good. For isolating memory problems, try swapping memory DIMMs with a co-worker's or friend's machine.
...
If you are overclocking then try increasing the core voltage, reduce the CPU speed, reduce the front side bus speed, or change the memory timings (CAS latency). Also try asking for help in one of the forums above - they may have other ideas to try."

So, according to prime95, errors could be caused by the memory crapping out, which could easily be the case for me since I'm using generic ddr 400 ram, so I wouldn't be surprised if even doing 200mhz was a strain on it.
 
Try running your RAM with a ratio so that the speed of the RAM is below 200 when the processor's FSB is higher. That way, the RAM is running within spec while you have the CPU overclocked. In most cases, this will allow you to see if the processor or the RAM is keeping you from a higher overclock.

I don't suggest running the RAM and FSB at anything other than 1:1 or 100% while you are actually using your system since that usually kills performance on AMD setups. Just use the ratio when trying to troubleshoot what is limiting your overclock.
 
Power, power, and more power. I've found psu's to be a large contributer to prime errors.
 
Originally posted by JackNapier
Power, power, and more power. I've found psu's to be a large contributer to prime errors.
Click to expand...

I just picked up a Seasonic PSU (reviews here http://www.silentpcreview.com/Sections+index-req-viewarticle-artid-107-page-1.html and here http://www.silentpcreview.com/Sections+index-req-viewarticle-artid-97-page-1.html) to address that very issue.. I am using some totally no-name craptacular 300W PSU right now that doesn't even have a switch on the back..

But anyway get this.. it doesn't happen at stock 1.83GHz but at 2.2GHz and higher... when I run prime95 torture test.. my voltages sag enough to hear the fan speeds in the system all drop :eek: :eek: .. It's really shocking to notice for the first time and to go and start-stop-start stop the test just to make sure you're not going crazy.. !

As you can guess I am not too lucky in the >2.2GHz region right now but that's all about to change :D Watch those PSUs people..
 
You also need to remember Prime95 is one of the few (only?) testing apps that actually verifies the answer your box is generating. When things run faster and hotter with lower signal to noise ratios you get bit errors. Many times the only way to resolve the problem is to lower the frequency. Unless you simply don't care if your machine generates the wrong answer occasionally, which for many people is true.
 
I have the exact same problem. I have an unlocked CPU and have played with the 10x multi, and I haven't been able to get consistent prime95 stability above 200.

Every now and then it won't even run stable at 200x11, so I'm suspecting that maybe my PSU is the culprit here.

With memtest86 I've tested my memory to be stable at 235mhz (235x10 :cool: ) with 2-2-2-11... but prime95 won't run :(

It sucks because I paid a LOT for this psu ($100 I think? 3 years ago). Enermax EG465AX rated for 433W with 38A/44A/33A on 3.3v/5v/12v... max combined 3.3v/5v is 220W.

I guess it's very possible that my CPU just won't do it, but I am keeping hope that it's something else. Maybe I'll order a new PSU today :p
 
When I o/c my CPU I noticed it was my ram causing the problem. And my PSU did take a small hit on 5v rail, it went from 4.93v to 4.88v underload while o/c. If I relaxed the memory timings the errors did not happen. No matter how much volts I used the error would still happen unless I slowed down the memory latency timings. Hope this helps and also monitor your rails while running Prime95 to see if they are dropping to much.
 
Since this thread brought it up, I decided to go back and try one last time, and raised my DDR voltage to 2.8v (from 2.7v) and my chipset to 1.7v (from 1.6v)...

My barton is set to the same as before, 1.75v.

Prime95 is now cruising for the last 2 hours at 220x10 :cool:

By the way, does anyone know which test is best to run in prime95? I'm running the blended, which tests a lot of RAM.
 
Well, blended is best if you know your RAM is working alright. If it isn't you could get errors because of the RAM, and not your CPU o/c. I would run memtest86 before using Prime95 for testing your CPU speed, especially if you are using blended.
 
I can sympathize, prime is driving me nuts, ran memtest86 all tests for 59 loops with no errors, can play BF for hours with no probs, Prime for 3 hours, then execution halted..I'm about to un/reinstall Prime95, corrupted install is about all I can hope for.
 
You must log in or register to reply here.
Back
Top