stupid question?

nicnac1 said:
hmmmm.... not quite the flat line vcore you got lol!
Click to expand...

Heh yeah you could say that! ;)

Your suffering from a bad case of vDroop and fluctuating vCore. Your chip gets more voltage when idle and less voltage at load (when it needs it most), the voltage it is getting is really unstable like its been drinking too much coffee! :D

2009090918h57vcore.png

2009090916h20vcore.png



I'm not an expert on this area but I believe there are several reasons this is happening:

  1. The PSU is flakey/overheating
  2. The PWM Hardware is low grade
  3. The PWM Hardware is over-heating

There are other reasons but those three spring to mind, the first two is a big problem but the third (overheating PWMs) can be remedied with improved cooling. The PWM's take the power from the PSU and convert it into processor friendly current, if the PWMs get toasty they don't function as good as they would if cool. In Theory if you can cool them down that vCore ripple should steady out, it's certainly worth trying! :cool:

nicnac1 said:
temps are lower than with prime95 (i assume that's because it doesn't torch the system to the same extent)!
Click to expand...
Yup! :)

See how it fairs with a large Data set! ;)
 
i hope then that it's the PWM overheating... i'll keep an eye on it during the next set of testing...

in the meantime i've tidied the inside of my case

Before

snv31215.jpg


After

snv31219.jpg


I've hooked some of the cables out of the way and moved the HDD lower down so it doesn't block any airflow... (this case is really not great for cable management)!

Idle temps are 5C lower already :)

I'll get cracking with a large data set...
 
just ran large data set... here are the results... still a crazy vcore!!

2009090920h33cpu1.png

2009090920h33vcore.png

2009090920h33pwm.png


don't think the temps are much better but i imagine a large data set would increase the temps anyway...
 
Last edited:
Good work with the case tidy, a 5°C drop (for free) is most welcome! :)

I'm wondering is OCCT showing you your PWM temps? If not then flick into the OCCT options and add the PWM if its not there already. Your board has a uGuru chip which shows all kinda of voltage and temp info so make sure OCCT is graphing that too!

if 3.4GHz (9x378) is passing multiple runs of OCCT then maybe switch it up to 3.4GHz (8x425) and see how you get on! :cool:
 
Last edited:
Big.Wayne said:
Good work with the case tidy, a 5°C drop (for free) is most welcome! :)

I'm wondering is OCCT showing you your PWM temps? If not then flick into the OCCT options and add the PWM if its not there already. Your board has a uGuru chip which shows all kinda of voltage and temp info so make sure OCCT is graphing that too!
Click to expand...

Just added it above

Big.Wayne said:
if 3.4GHz (9x378) is passing multiple runs of OCCT then maybe switch it up to 3.4GHz (9x425) and see how you get on! :cool:
Click to expand...

do you mean 8x425?
 
I do indeed mean 3.4GHz (8x425) ;)

Can you also try the other CPU test from OCCT (CPU:Linpack), that should heat things up a lot more, hopefully it doesn't crash!
 
Last edited:
Excellent!

When you originally fitted that gargantuan HS did you use the pre-applied paste?

If you did - and even if you didn't - as you're taking it all apart you may want to consider reseating the heatsink with some quality TIM (MX2/3 or AS 5) – just to rule out *slight* poor CPU/HS contact as an added possible reason for your extreme temps - coupled with poor case circulation.

I use the pea sized blob method and let the weight of the HS and heat do the rest - as i've noticed this method reduces the risks of air getting trapped (i used to use the cling film method and spread it evenly over the core - it's been a hard habit to break).

With a quad core you're meant to apply a small line down the middle of the CPU - but i've found that the blob method still works.

MX2 instructions – link

Artic Silver instructions – link
 
Last edited:
Hi Plec,

Yeah I used Artic Silver 5 when I first set it.

I used the "thin line" method , perhaps i didn't put enough on (still lots left in the syringe), or there's air trapped or something. Either way this time i'm expecting to see a decent improvement next time.

Some people are recommending aiming the heatsink at the top fan although i'm not sure what that does for the circulation or even if there are any benefits.

I've heard that the best method is to have the 2 tricool fans on low and the intake fans on high to get the best results... I imagine there will be a lot of tweaking to find out :)
 
nicnac1 said:
I used the "thin line" method , perhaps i didn't put enough on (still lots left in the syringe), or there's air trapped or something.
Click to expand...

Sounds as though you used exactly the right amount - 'a small thin line'- you only need enough to spread and fill any minute imperfections/gaps where the 2 surfaces meet.

nicnac1 said:
Some people are recommending aiming the heatsink at the top fan although i'm not sure what that does for the circulation or even if there are any benefits.
Click to expand...

Well, as your well aware, heat rises and the top exhaust would take the vented heat from the HS straight out the top – if it was orientated that way i.e. fan facing the gfx card.

However, if you had the HS directed towards the rear outtake - the excess heat that rises from the HS would still vent out the top – plus, you would have additional ‘coolish’ air flowing over PWM from the HS fan… (personally I prefer the latter method – but again it’s open to debate and I would follow BigWayne’s advice as he uses this case – I think - and will be in a better position to advise)


nicnac1 said:
I've heard that the best method is to have the 2 tricool fans on low and the intake fans on high to get the best results... I imagine there will be a lot of tweaking to find out
Click to expand...

Hehe, - now there’s a thread title that could cause a small apocalyptic war - positive or negative pressure?

I’m a middle of the road man and have all my fans as slow/quiet as I can get away with – intakes at the front – outtakes at the rear/top – simple but works for me.
 
Last edited:
Evening all...

here are the results of OCCT with the chip running at 425x8 1:1

2009091020h47cpu1.png

2009091020h47pwm.png

2009091020h47vcore.png


Can't wait for the new case tomorrow :D:D Going to play with the fans to see which is the best setup :) maybe end this war! lol
 
Hey nicnac1,

sorry I didn't get to update you tonight but it looks like everything is moving along nicely! :)

If you get time can you do a OCCT run with your BIOS set at full defaults please, I am curious to see how the CPU/PWM temps are when the chip is running bone stock but I mainly wanna the how the vCore load is with the chip not overclocked . . .

Hope you like the Antec 300! :cool:
 
will put it back to stock and test after work tonight. Will post the graphs again.

fans have arrived... no sign of the case yet (although admittedly i didn't order it from OCUK)

can't wait! :D
 
2.4GHz (9x266)

36°C Idle @ 1.305vCore (39°C PWM)
58°C Load @ 1.2705vCore (54°C PWM)

I was expecting those temps to be lower and the vCore ripple to be slighty less! :p

Have you done any Undervolting on that chip yet? With all the other voltages on [Auto] start dropping the vCore down, if the chip doesn't get enough juice your system will reboot, be brave and dial it down to 1.200vCore or lower and try and get off a ICCT test when you think you found out its roght minimum.

Undervolted that chip should run cooler, the PWM's should run cooler and that vCore ripple should flatline a bit more. This isn't a full time scenario but more testing purposes! :cool:
 
ok... will drop the vcore as much as possible and let you know what happens
i imagine the temps will be better when the new case finally arrives??
 
Heh oh well, some boards allow you to set voltage lower some don't?

So if you set the vCore to the lowest possible and then overclock the chip to 3GHz (333x9) does it still run stable?
 
You must log in or register to reply here.
Back
Top Bottom