Winnie on Water. Less OC?

EclipseJP · Feb 11, 2005

I had aquired my TDX block and had a BIX radiator and 1048 from Previous water setup. I had gotten to 2.7 with my 0449 on the stock heatsink with ease, within the first hour of getting it.

Since then I have added a OCZ Powerstream 520 and the 1GB G.Skill 4400 LEs and also put some heatsinks on the mosfets. Everything you need to ensure a good overclock.

Now after putting on the water she will not budge past 292 fsb. :temper:

I have tried with setting the memory divider down to 133.

I have been experimenting with adding more voltage and it just wont go anymore. She'll do 2.56 on stock voltage. 2.63 with +.125 but just does not want to go any higher. I am not sure that it has to do with the waterblock not completely covering the heatspreader, but the cpu is staying at 31 idle 38 load (the AS5 has yet to set in) at the 2.6 w/+.125.

I was really hoping for a 2.8 overclock, but I dont think thats gunna happen.

Perhaps the WB is on too tight and pushing down is bad? Any ideas?

sandman001 · Feb 11, 2005

I'd try reseating it, and if that doesn't work try out the stock heatsink again?

EclipseJP · Feb 11, 2005

I suppose I could try that.

Also does the fact that I went from 2 256 sticks to 2 512 have any effect? I was running the 2 256 ones at 133 memory ratio.

I am also going to try to do single channel first before I reseat the WB. Maybe its just my board not liking anything. I see that a certain member is having great luck with the DFI Lan Party.

deception`` · Feb 11, 2005

EclipseJP said:
Also does the fact that I went from 2 256 sticks to 2 512 have any effect? I was running the 2 256 ones at 133 memory ratio.

It sure does. Typically, we find that it's harder to overclock using 512 sticks compared to smaller, single-sided modules. Of course, the K8 platform is not as sensitive to this as an NF2, but it's still a common issue.

deception``

EclipseJP · Feb 11, 2005

deception`` said:
It sure does. Typically, we find that it's harder to overclock using 512 sticks compared to smaller, single-sided modules. Of course, the K8 platform is not as sensitive to this as an NF2, but it's still a common issue.

deception``

Hmm. Is this an issue no matter what speed you run the memory at?

deception`` · Feb 11, 2005

EclipseJP said:
Hmm. Is this an issue no matter what speed you run the memory at?

It has to do with the fact that 512 MB sticks are double-sided. It doesn't make it impossible to overclock like this, but it does mean that running 256 modules is a tad easier. Also, some architectures are more sensitive to this than others, and the K8 platform tends to play nice with double-sided DIMMS. So you'll only lose a couple HTT/Mhz on average.

deception``

EclipseJP · Feb 11, 2005

Ok thanks for clearing things up.

JDizzle · Feb 11, 2005

Run memtest and find out what's the max for your ram. You said you're running a 133 memory divider so I don't think there's an issue with the ram and it being 512mb instead of 256mb.

deception`` · Feb 11, 2005

JDizzle said:
Run memtest and find out what's the max for your ram. You said you're running a 133 memory divider so I don't think there's an issue with the ram and it being 512mb instead of 256mb.

As I said, the use of a divider is irrelevant. It's simply the fact that he's running double sided sticks (it stresses the memory controller more).

deception``

JDizzle · Feb 11, 2005

So would running one stick of his 512mb sticks produce the same results as before?

samuraisam · Feb 11, 2005

Well, for the max overclock you're going to have to stick with the two 256 sticks, 512 max amount of memory for that oc.

-Sam

EclipseJP · Feb 12, 2005

OK. I figured it out. Its indeed my processor. Its just not a very good one. I do 2.62 no problem with 1.5v. But when I want to go above the 292 barrier, it requires no less then 1.65v. I really dont like running that through it. The LEs were in when I got her to 2.7 again, but she crashed after about 5 minutes in windows.

I suppose I just didnt get a good one. O well with Venice hopefully around the corner, maybe that will be an option. Till then I will just run her stock and sell it off when the time comes.

yeha · Feb 12, 2005

just wondering, how much airflow are you getting around the cpu socket? i see you put heatsinks on the mosfets, but unless you still have some airflow to cool the sinks they're only going to be buying time.

perhaps try mounting an 80/92mm fan over the socket area and seeing what happens. or stick your finger on those heatsinks while you're running ~2.6ghz and see if they leave a burn mark

just an idea.

Jhatfie · Feb 12, 2005

EclipseJP said:
OK. I figured it out. Its indeed my processor. Its just not a very good one. I do 2.62 no problem with 1.5v. But when I want to go above the 292 barrier, it requires no less then 1.65v. I really dont like running that through it. The LEs were in when I got her to 2.7 again, but she crashed after about 5 minutes in windows.

I just got a DTR 3000+ that has similar traits as your winnie. It can do 2.66ghz at 1.55v stable, 2.7ghz stable at 1.65v and basically craps out much after that no matter what the voltage. I can boot at over 2800mhz and bench at 2750, but it is not what I would call stable even with 1.75v. Memory dividers, HT multi's, chipset voltage and so on make no difference. I am running 2x512MB TCCD's, but using one did not make a difference. Still, I can't complain.

Lifthanger · Feb 12, 2005

JDizzle said:
So would running one stick of his 512mb sticks produce the same results as before?

No because there are two memory controllers. Running one doublesided stick would still stress one of these controllers, while the other one is idle.
They can't split up the work, because they have their own "private" connection to the memory banks (the famous dual channels; one for each controller).
Maybe less heat is generated when only one is stressed, but I don't think it would make a difference.

EclipseJP · Feb 12, 2005

I have 2 80mm fans mounted on the case window, one blowing right towards the socket and the other below the vid card, and the front I have a 120 sucking air through a radiator, and of course one in the back of the case sucking out.

But maybe I shall put a 60 mm fan on the mosfets. I will see what I can muster up as far as fans go.

EclipseJP · Feb 12, 2005

Jhatfie said:
I just got a DTR 3000+ that has similar traits as your winnie. It can do 2.66ghz at 1.55v stable, 2.7ghz stable at 1.65v and basically craps out much after that no matter what the voltage. I can boot at over 2800mhz and bench at 2750, but it is not what I would call stable even with 1.75v. Memory dividers, HT multi's, chipset voltage and so on make no difference. I am running 2x512MB TCCD's, but using one did not make a difference. Still, I can't complain.

I feel your pain, but mine is not even stable at 2.7 with 1.65v. Mine is a newer chip too (0449) and thats what I really dont get.

piotrr · Feb 12, 2005

You really need that voltage? My Winnie rather seems MORE unstable the more juice I feed her.

EclipseJP · Feb 13, 2005

Ya she needs that voltage. It just wont post and I have to clear the CMOS if I don't give it that much juice above 293 fsb mark.

Rumrunner · Feb 14, 2005

Few people see a better oc than that with winnie. Just think, those are faster than stock fx-55 speeds. $825

Winnie on Water. Less OC?

Member

Just Freeze It

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Similar threads