Help - Unstable System

M

m4tthew

n00b
Joined
May 22, 2007
Messages
28
Before i start this is my system

Antec 900
q6600 G0 Stepping
Gigabyte EP43-DS3L
Xigmatek HDT-s1283
Corsair XMS2 2x2GB
ATI HD 3850




Ive managed to overclock my cpu to boot to windows at 3.00ghz but one of the cores fails prime95's torture test after between 5-9 hours. Ive set all my motherboard settings manually to the "auto" vaules such as Core Voltage, core speed, memory timings, etc etc.

Having tested it a few times and it still coming up with this error i tried upping the core voltage to see if that affected anything. However instead of core 1 failing, core 2 now failed.

Im not sure where to go from here... RAM? I ran memtest but it said they were fine.:confused:
 
Make sure you have the energy saving stuff disabled, the is a good thread on here about gigabyte boards for overclocking them.
 
yeah all the energy saving stuff is turned off.

PS- why is the cpu-z clock speed different to my bios setting?
 
CPUz reports actual clockspeed, whereas BIOs post usually gives a ballpark or an estimated speed. Being slower/faster by a few Mhz is normal.

Try setting your vcore to something like 1.35 or 1.38. Even though I can leave my vcore on auto up to 3.2GHz (same processor), Everest reports that she draws about 1.38-1.39v at 3.2Ghz operation.
 
ok ill try upping the vcore incrementally and see if it stablises it.

Just curious... is it completely random which core fails under prime95?
 
Sorry... do you mean setting the vcore in bios to 1.35 or what it is rated at in cpu-z ? Which is the one that people qoute when in the forums?

Im guessing that the one in cpu-z is with Vdroop but is this the figure people use when overclocking?

New to overclocking :confused:
 
Hello, I have a problem very similar to what matt has. One of my cores will fail in P95 blend test after 2-3 hours every time. However, the other core keeps going without errors after about 6-7 hours before I stop it. I thought maybe it could be the RAM so I booted up the latest version of memtest86+ and it had no errors.. I only ran it twice though.

Last night I ran the "small-fft" test or whatever it's called (where it mainly stresses the cpu/fsb) for 7-8 hours and go no errors on either core.

The ONLY thing I have overclocked is my cpu. My RAM and mobo support the fsb I'm currently at: 1600mhz. I really don't believe I have any faulty hardware and maybe a voltage is too low/high because I never freeze up or crash. I can play any games I own just fine. I have gotten a few BSoD's but it's been awhile and that was with different timings, when memtest86+ DID report errors but OCZ helped me resolve those errors.

I'm just wondering why one of my core fails in the blend test after 2-3 hours but the other core can keep going?

Current settings:

cpu voltage: 1.3v (problem happens with 1.315v and 1.325v as well)
fsb voltage: 1.25v
northbridge: 1.375v

7-6-6-24 timings @ 1.9v (currently testing at 1.86v)
trfc: 72
trrd: 4
twr: 10
twtr: 5
command rate: 2t
These settings should be supported by my ram.

I know people have problems OCing on my mobo, but I was able to boot and operate when I had the fsb at 430. I don't want to risk going further because it gets a little hot.. Mainly because my PC is in a pretty hot room but I have good airflow and cpu idles at 40-45c and full load after an hour its 60-65c at current settings.

I'm thinking I may just need to adjust my fsb or nb up more... But I've already done it a few times. I didn't think I'd have to add too much voltage to OC from a 333 fsb to 400.... Any suggestions?
 
The threads that fail in Prime95 are the ones most susceptable (or less tolerant) to the settings you currently have. The thread (core) could change at any time given different core voltages changes, temps, you name it.

Just go into the bios, and try incrementally raising the vcore (CPU Voltage) to something in the ballpark of 1.35v-1.4v and monitor stability. You shouldn't require more than that.
 
Thanks for the advice. So far what I did is drop my vdimm from 1.9 to 1.86 and have been running Prime95 for almost 5 hours and no errors on either core so far. I'm going to continue to let it run. Hope it passes my expectations. 62c temp on core in my hot ass room. Not too shabby.

If anything fails I'll keep trying I guess.
 
okay so i have got it stable at 3.0ghz ... tried 3.2ghz and it will boot into windows and when i start prime95 it stopped one of the cores and froze on me, had to reset. So i upped the voltage to see if that would stablelise it. Ran prime95 again at a higher voltage and no cores fail but comp just freezes...

Any ideas?
 
just wondering where people get their figures from ... i can imagine myself setting the vcore ridiculously high because i thought people were using figures in cpu-z rather than bios :)
 
so freezing whilst running prime95 is just normal ? Everything is fine for about 30 secs and then windows freezes. No BSoD or errors...
 
m4tthew said:
so freezing whilst running prime95 is just normal ? Everything is fine for about 30 secs and then windows freezes. No BSoD or errors...
Click to expand...

Of course it's not normal. What are all your current voltages? Something may be too low if it just freezes. Does it happen all the time when running p95?

Anyways, I'm still have troubles as well. Sometimes I wonder if P95 is random. I decided to stop P95 last night after running it for ~6 hours with no errors and wanted to do some things on the computer. So eventually I started playing Bioshock and everything was doing fine, then I started noticing slight flickering occurring in the game. I minimized the game and then got a BSoD :(

Started doing more testing and P95 failed on the first core after 3.5 hours. I had to bump my vdimm back up to 1.9v and bumped my northbridge up one more at like 1.37v.. Currently testing, we'll see how it works.

I wish there was something in P95 that gave you a better idea of what to look for when receiving an error :p
 
You must log in or register to reply here.
Back
Top