possible prime95 fault? what's your experience?

[MoonPie]

ok this is my setup
xp2100 , asus a7n8x dlx. rev. 2 , hyperX pc3000 , sk-7

been runing prime95 at 191 FSB , mult 11 , vcore 1.6 (agp freguency set to auto) and no errors for 5 hrs...

so i decided to up fsb to 196 then i got error right off from prime95 so i went back to 191 fsb and tested again....still gave me error...

is this prime95's glitch? should i not worry about it too much on prime95 results?

 

[Amorphus]

errors = t3h suck. if you're doing prime95 for actual distibuted computing, you should worry about it, but if you're only doing it for the benchmarks, then no problems, unless you're experiencing problems.

and the xp2100+'s default vcore is 1.6... maybe you should up it to 1.65?

 

[Duvie]

You ppl need to remember this software when ran in a torture test is stressing the system to find weaknesses....unfortunately they can damage components like any stress testing can....I would take prime95 as serious sice it was working before. I would see if anything else has changed. How where the temps/?? how were the voltages???

You are stressing power supplies, ram, mobo chipsets, cpu, etc when stressing....

 

[MoonPie]

yep vcore is 1.65
temp load 40 / idle 35 currently using the lite-on ps 300w

just can't figure it out all i did was change back to when it was stable at 191....

 

[THUGSROOK]

thats all it takes to hurt an AMD.

not stable @ 191fsb no more? then you hurt the chip.

you must be able to pass 5 hours of p95 or something is wrong with your system.

 

[ianching]

5 hours doesn't gurantee what I call rock solid. My 1700+ at 2.2GHz at 1.625V fails Prime95 only after 9hrs (runs fine at 1.65V for 24hrs+).

I would say running 24hrs to ensure the stability.

 

[bgeh]

Originally posted by: MoonPie
ok this is my setup
xp2100 , asus a7n8x dlx. rev. 2 , hyperX pc3000 , sk-7

been runing prime95 at 191 FSB , mult 11 , vcore 1.6 (agp freguency set to auto) and no errors for 5 hrs...

so i decided to up fsb to 196 then i got error right off from prime95 so i went back to 191 fsb and tested again....still gave me error...

is this prime95's glitch? should i not worry about it too much on prime95 results?

what's your current config?
is it 12 x 191?
or 11.5 x 191?

 

[MoonPie]

now currently 186 FSB multi 11, and vcore 1.625

anything abover 190 fsb seems unstable
temp now 43 load...

just can't hit 2.1 Ghz

 

[GUN]

What OS?

 

[onza]

i really dont understand what the difference between 23hrs and 24 hrs of testing will do.. i mean if it poops on you i'm sure it'll do it in the first few hrs..

does prime get more extensive as the hours go by?? or is it a loop system that does the same iterations over and over again?

onza

 

[ChampionAtTufshop]

Originally posted by: onza
i really dont understand what the difference between 23hrs and 24 hrs of testing will do.. i mean if it poops on you i'm sure it'll do it in the first few hrs..

does prime get more extensive as the hours go by?? or is it a loop system that does the same iterations over and over again?

onza

afaik, its a loop system

thing of it as running across a shoddy bridge
the first few, even maybe 10-20 times, of runing across it wont destroy it ...probably give minimal damage..like swaying back and forth, but nothing critical
but when u get into running across it maybe 200 times non stop, you have a good chance of faling to your doom

things fail under prolonged periods of [extreme] stress.
like those 24hr lemans races or wahtever kind they are...you dont want a car craping out completely at 23:59:30 do ya lol

my stress test includes running a machine for a week or two continusoulsy with prime95 in the background, while doing what i would normally do

 

[Tanclearas]

If you read up on Prime95, any errors that appear are due to your system in one way or another. Prime95 calculates the values (using a stable algorithm) and compares them to known correct answers. As for how long it can run before crapping out, I actually had it fail after 36 hours of torture testing.

 

[Duvie]

Thugs is pretty much right for most ppl here....Unless you are doing distributed computing the system really needs to be stable for periods that are common to your use....I usually go 6-12 hours....

As for the newer p4 ppl (not amd's as of yet) we need to look beyond prime95 as it has been shown to be unreliable in detecting problems. When it detects them they are there but there are some times it does not...I have had 12 hours prime95 stable to have it fail in regular apps in 10min to 1 hour...vcore boost solved it but stability needs to really include the programs in which you run....no errors are exceptable!!!

 

[Naruto]

I'm surprised my setup could run 3dmark 2k1 loop and 1/2 hour of mohaa with full detail on. Yet it failed the very first test of prime. I lowered the fsb by a simple 2 mhz and boom, prime was stable.