prime95 problems..

[-Blue-]

hey all.. for some reason i decided to run prime95 to see how my OC was doing... not that i had, had any problems while running anything or crashes but just for kicks and to see how hot it got..

core0 runs fine on all 3 tests but core1 keeps crashing.. on the 2 big tests almost instantly while on the small one after test4 or so.. and i don't get it.. i've upped the voltage a bit to see if that got it kicking but nothing.. and i'm curious why...

the error is the got .5 when expected .4

temp on both seemed fine.. 54C max on both, ambient is 28... any other info you should need let me know and i'll add it.. if it's BIOS info there are a lot of things i have no clue about to be honest.

only voltages i touched were CPU and RAM.. most of the other ones aside from NB/SB i don't have a clue what they mean or are..

well any help is welcomed..
thanks in advance

 

[Evilsizer]

are you running blend,small or large? if running blend/large look at the size the errors are occuring. are they happening at say 1024k, as this will tell you were the error lies. if you have run small TFT's and the cores do not fail then its not ur cpu. it could then be either your ram or needing more voltage to the NB. i would suggest getting memtest86 if prime95 is not failing on small TFT's.

 

[-Blue-]

i'll let you kow as soon as i can.. but i do think it was at 1024K

 

[Evilsizer]

ok well if anything fails in 0k-4096k =cpu since the E6600 has 4mb L2. if it fails over that, just like i said above.... you are also running 4x2gig sticks,so add more NB voltage....

 

[-Blue-]

good to know that.. here is the screenie..

is there any way to know what the current voltage on the NB is? since i have it in auto i have no clue what it is now

 

[Evilsizer]

well i can tell you its the cpu, as cpuz is telling you,your cpuv is 1.264v. in bios up it more till you see it read 1.3-1.35v. that is just a range as i bet 3.2ghz will be stable with 1.3 and you may be able to do 3.4ghz with 1.35v.

as for checking voltages, while in bios, check the monitoring section. other wise go manual, set it to 1.3 or 1.33. its some weird voltage option like that not like ,1.3/1.31/1.32, but bigger jumps.

 

[-Blue-]

ok i upped both cpu and nb voltage..

cpu on BIOS = 1.365 -> cpuz = 1.32 (drops to 1.296/1.304 when on full load)

and NB i had 1.25, 1.4, 1.55, 1.7 as options and it's at 1.55 right now..

this time around it small FFT's and in-place large FFT's passed the first set of 11 tests
blend however still fails...

so far it's a step forward since the large one used to fail too..

 

[Evilsizer]

run memtest86 then if its failing past 4096k,also you dont need that much NB voltage, all you really needed was 1.33ish,so set to 1.4,then take the cpuv up one more notch then...

 

[-Blue-]

i dropped the NB to 1.4 and upped the cpu by 1 setting so it's now at 1.328
i also put the RAM at it's default settings since i had changed them from 5-5-5-15 @ 1.9 to 6-6-6-18 @ 1.8

i ran the blend test and it went fine.. i stopped it after almost 4 hours and no problems
i think it might have been the RAM

 

[SkiBum1207]

yeah I would reccommend for overclocking, set the FSB u want, and then set the Ram to a high latency/low speed. That way you can pretty much guarentee any errors are the CPU, once you get a stable CPU, then go for RAM, because you know the CPU is stable, you can then tweak the ram. Also, You could pencil Mod the board, I know with my P5k vanilla the main hinderance in OC was the Vdroop, check how much the voltage drops when you load the processor vs idle Voltage

 

[-Blue-]

yeah, i just thought that if i upped the voltage to 1.9 on the RAM from the 1.8 original i would be able to have them running i set down.. but i guess not...

lets see how i do with the Q6600 that in theory arrives either this afternoon or tomorrow morning, i thought it would be good to first try on that quad before going to the Q9450.. since i was planning on building a micro system with a Q6600.. i'm sure i will be having to ask for help around here again