Overclock.net banner

1 - 16 of 16 Posts

·
Registered
Joined
·
108 Posts
Discussion Starter #1
Tried running Prime95 and immediately got an error message about "Lucas-Lehmer iterations" (whatever they are):

"Fatal Error: Rounding was 0.4520950317, expected less than 0.4. Hardware failure dectected."

What is that about? Computer is very stable. I just ran 10 cycles of Sandra 2004 without any problems. Tested ram recently with MemTest86 and it checked out fine.
 

·
Premium Member
Joined
·
11,430 Posts
Prime95 is extremely picky and will hault on the the smallest of errors. That basically means your processor isn't stable by its measure (sandra isn't a very good measure either, I did over 200 passes with it and still had crashes).
 

·
Premium Member
Joined
·
11,430 Posts
Too much heat can cause system instability as well, that error could be the result of an overheated system also
 

·
Premium Member
Joined
·
11,430 Posts
Got any other detailed info? The program istelf is far from perfect too, maybe try reinstalling it
 

·
Registered
Joined
·
108 Posts
Discussion Starter #8
Well, I don't think heat is the issue but I could be wrong. Processor is running at 42 C idle and about 50 C when Prime95 is loading it (I acutally got it to run Prime95 for 7 minutes last night after changing a bios dram setting). MB/case temp is being reported at 31 C idle. Are those temps too high?

No more info than what I already gave as far as the Prime95 error message.
 

·
Premium Member
Joined
·
7,395 Posts
Quote:


Originally Posted by trents

Well, I don't think heat is the issue but I could be wrong. Processor is running at 42 C idle and about 50 C when Prime95 is loading it (I acutally got it to run Prime95 for 7 minutes last night after changing a bios dram setting). MB/case temp is being reported at 31 C idle. Are those temps too high?

Your temps are comparable to mine when I run Prime95, and mine runs Prime stably indefinitely.

If you turn down your OC, can you run Prime stable? If so, then you've found your problem. The question then becomes whether you are willing to run at a slower speed and be rock solid stable, or prefer to keep your extra bit of speed and be potentially a little unstable.
 

·
Registered
Joined
·
108 Posts
Discussion Starter #10
Well, I returned the system to stock speeds and voltages and it ran Prime95 without a problem. Now I'm going to manipulate settings to try to isolate whether its the cpu or the memory that is causing Prime95 failure under overclocking conditions. I figure if I lower clock cycles while increasing fsb to keep processor at stock speeds I can check for ram weakness. The inverse should test for lack of headroom with the processor.
 

·
Premium Member
Joined
·
4,555 Posts
Quote:


Originally Posted by trents

Well, I returned the system to stock speeds and voltages and it ran Prime95 without a problem. Now I'm going to manipulate settings to try to isolate whether its the cpu or the memory that is causing Prime95 failure under overclocking conditions. I figure if I lower clock cycles while increasing fsb to keep processor at stock speeds I can check for ram weakness. The inverse should test for lack of headroom with the processor.

The other day my computer primed to close to 24hrs without a problem... Now I'm getting this same error... Oh well, I say screw prime95.

I havent experienced a single crash, I fold 24/7, I play games while folding ... Stable enough for me.
 

·
Registered
Joined
·
108 Posts
Discussion Starter #12
Yea, I can run a lot of demanding stuff also and no crashes when I am overclocked to 11.5x200 with core voltage at 1.825. I can run Sandra 2004 burn in plus other stress testers, AquaMark 3, Unreal Turnament, ect. without any problems. Only with Prime95 do I get problems. Another interesting tid bit is that my idle speed core temp is 42C whether I'm running the cpu at under 2.0 ghz. or overclocked to over 2.3 ghz. The load temp rises higher however, when I'm overclocked, say 50C as opposed to 44C.
 

·
Premium Member
Joined
·
4,555 Posts
Quote:


Originally Posted by trents

Yea, I can run a lot of demanding stuff also and no crashes when I am overclocked to 11.5x200 with core voltage at 1.825. I can run Sandra 2004 burn in plus other stress testers, AquaMark 3, Unreal Turnament, ect. without any problems. Only with Prime95 do I get problems. Another interesting tid bit is that my idle speed core temp is 42C whether I'm running the cpu at under 2.0 ghz. or overclocked to over 2.3 ghz. The load temp rises higher however, when I'm overclocked, say 50C as opposed to 44C.

I raised my voltage to 1.9v and the errors didnt come on like it did before.. I think I primed to like 10 mins and stopped it.

However, I'm not confortable using this high voltage so I backed down to 1.75
 

·
Registered
Joined
·
125 Posts
Speaking from experience....I've seen almost all types of p95 errors. The one you got is probably caused from a lack of power to either your processor or memory. Try bumping up your vcore or your vdimm, that should help. I was getting the same rounding error, and then I bumped up the vcore and now I can run prime95 for over 8 hours with no errors.
 

·
Registered
Joined
·
108 Posts
Discussion Starter #15
I think I've narrowed it down to the memory being the problem. When I run it at 166 mhz fsb instead of 200 mhz it does fine with Prime95. It's only Kingston Value ram so I can't expect much. Tomorrow I'm going to ask the forum experts for DRAM timing relaxation suggestions.
 

·
Registered
Joined
·
18 Posts
I'm still experiencing problems, that rounding problem and also another one, stating that the hardware failed.

My system is still running on stock speeds, nothing is OC'ed. Only case and cpu ventilation are radicaly modified. Cpu temps never rise above 45° now.

Next week, I receive a new psu, maybe it'll be better...
 
1 - 16 of 16 Posts
Top