Winnies don't max at 1.55V... do they?

E

ecxeleron

Weaksauce
Joined
Jul 3, 2003
Messages
77
I have a winchester 3000+ in an ASUS A8N Deluxe.
I have had prime and hl2 stable operation at 300x8.5 1.55V (2.55GHz) I just managed to bood into windows at 2.7GHz (300x9 1.55V)
when i vmodded my board to have a vcore of 1.66V, however i had the same unstable-ness at 2.7GHz as i have had before the vmod.
Is my chip maxing?

I have watercooling

I did the fixed vcore vmod and the chipset vmod (currently set to 1.67V) on this page http://www.vr-zone.com.sg/?i=1624&s=1
 
from what i've seen it seems that vcore doesn't make that much diff on these processors. I had no idea that asus board could do that high a FSB. I'm getting my new epox 939 motherboard today , along with a 3000+ and a thermalright xp-90. Its going to have to oc fsb to 300mhz+ to hit over 2.7ghz... crud i hope the board can hang. honestly i'd be happy with any final clockspeed hopefully above 2.3-2.4ghz, so i can say to my wife "See i saved 800$ by overclocking"
 
it's rather likely that the chip isn't responding well to the voltage, and the transistors are being saturated, thus creating switching errors that you see as instability.
and besides, even on my clawhammer, i see very little gains for going over 1.55v
 
could be a number of things.

You're putting some major power out at 1.67/2.7 ghz, (by CPU power's calculations about 143W) the PSU might not be up to the task. (The motherboard voltage regulator may also be an issue, check those temps, if those mosfets are pushing 60 or 70C at load (at 2.55V), some cooling on them may be in order).

As Eclipse said, it's possible there some physical limit with the transistors, they don't scale forever, at some voltage, the channel becomes saturated, and pushing higher won't result in signifigant increaeses in drive current. It's also possible the added heat from the higher voltage is offsetting any real gain the voltage is making.
 
my PSU is up to the job, it's a toppower silent 450W modded with a faster fan. 3.3V rail has been modded to 3.4V
 
I've played with a couple now, and voltage doesn't seem to help much, I am 100% prime stable at 2.651 and 1.45 V's, however any speed above this independent of voltage is unstable (prime errors about 5 mins in). On the other hand I can drop the voltage down to 1.25 V's and run 2.4 ghz 100% stable. I also tried another winnie and had basically the same results except for max speed was like 2.55 ghz. I think what is happening is AMD actually got the 90 nm die shrink right (unlike intel) and therefore the processors clock higher and require less voltage. So usually at default voltage you can OC with about + 0.2 V and then it reaches a point where voltage doesn't help much. I think the same is true here its just that default voltage should really be about 1.3 V for the 90 nM processors
 
perhaps amd's yeilds are better than expected.

I reduced my vcore to 1.456V (2.55GHz) and i get an error in prime instantainously.
now running prime at 1.504V

Is it worth it buyiong another chip in the home of 2.7GHz? I have the necesary cooling for it.
 
ecxeleron said:
perhaps amd's yeilds are better than expected.

I reduced my vcore to 1.456V (2.55GHz) and i get an error in prime instantainously.
now running prime at 1.504V

Is it worth it buyiong another chip in the home of 2.7GHz? I have the necesary cooling for it.
Click to expand...

It wouldn't be worth it. You may get a chip that overclocks even worse then the one you have now or just simply wont hit 2.7Ghz. Most top out around 2.4-2.6Ghz.

Outside of benchmarks your not going to notice a difference between 2.55Ghz and 2.7Ghz anyways.
 
I don't know....i've got mine @ 2.5 on the stock voltage and the stock HSF...100% stable and the temps are around 45...might try higher when i get new ram and better HSF in a couple weeks.
 
why not just drop the ram divider and watch the temps carefully in the meantime? :D
 
The whole ram divider thing confuses me...is there or is there not an issue associated with running the ram and the divider. And as for the temps...what would you say is unsafe? higher than 60?
 
read the link in my sig for how the ram works..
and 200*11 with ram 1:1 will give results that will be almost exactly the same at 245*9 with the ram set to "166mhz" (5:6)
in other words, the divider doesn't effect performance because with the memory controller running at cpu speeds, it's never running 'in sync'

and yeah, 60c is a good limit to your temps. it's ok to go a bit over, but only if you're not overvolting your cpu.
 
stock voltage, whoa!
when amd comes out with the ss3 revision,i think i'll get another 3000+
 
You must log in or register to reply here.
Back
Top