Official DFI NF4Ultra-D/SLi owners thread.

ricco said:
2725mhz stable with 1.4v
2800mhz stable with 1.425v
2850,hz stable with 1.525v
2900mhz stable with 1.55v
2950mhz stable with 1.65v
No stability over 2950 possible even upto 1.8v
Click to expand...
Are these your results with the IHS removed?

I take it you have one of the SPMW ones as well?


I'm trying to see if I can find anyone that might have a TPAW. I doubt the OCing potential would be too much different from the SPMW but it would be nice to see some results for it :)
 
[timko] said:
Are these your results with the IHS removed?

I take it you have one of the SPMW ones as well?


I'm trying to see if I can find anyone that might have a TPAW. I doubt the OCing potential would be too much different from the SPMW but it would be nice to see some results for it :)
Click to expand...

I quoted my full code in previous posts, but here it is again.
CABGE0515SPMW
lot D50170

A different stepping result would be interesting.

And yes those are with IHS removed and watercooled, although I need to strip and check I have good core contact and maybe change to AS3(from As5) as I always have had better/consistant results with AS3.

I do think that 3Ghz is possible as I can run a whole world of benches, but its not stable (as yet)
 
I'm very tempted to jump the gun and grab a TPAW myself, I've already got two confirmed points of sale. I'd be sticking it under watercooling too but I'd be too scared to take the IHS off :(

Is it more a case of the IHS "traps" in some of the heat or that the contact between the core and the IHS isn't as good because AMD use the same kind of generic paste that HSF makers throw in for free?
 
Its a little bit of a combination.
1: IHS is not perfectly flat - infact i would go as far as to say its inferior to the STD AMD HSF. Generaly they are Concaved (from what I have seen lapping them), so they have higher edges. This is not good and dispells an ytheory about pressure used to get a better contact against core etc.
2: The TIM under the IHS has a thickness to it, might be only 0.1mm but its not good, nor is the fact its seems cheap stuff
3: Its just another thermal junction , and junction acts as a barrier and requires a slightly higher differencial to make the heat pass, the less we have the better we cool.

It all goes back to the Celeron days realy, we used to pop the caps back then and direct cool , then we did it with the P4 (except m0's) and now with the AMD.
It does stop many problems with core crushing for many, but waterblocks dont excert as much force on the Die as a HSF clip, nor does it require a side loaded clip.(with some exceptions)
I dont even use springs, i just clamp down. Always have done.
 
but does it really make that much difference off?you're o/c are pretty much same as mine and Concorde and he is aircooled.I also think phase change would show true voltage scaling with these.but im hopefull with some burn in the 3GHz barrier is possible.

so ive setup prime95 as max heat/power consumption(100% cpu) but custom so 1Gig mem is tested too.Going to run this overnight so should serve to burn in cpu and VX. ;)
 
Last edited:
monkeygabriel said:
but does it really make that much difference off?you're o/c are pretty much same as mine and Concorde and he is aircooled.I also think phase change would show true voltage scaling with these.but im hopefull with some burn in the 3GHz barrier is possible.

so ive setup prime95 as max heat/power consumption(100% cpu) but custom so 1Gig mem is tested too.Going to run this overnight so should serve to burn in cpu and VX. ;)
Click to expand...


Well you say that, but you 2 might have slightly better CPU's then me to start with. Rem not every IHS will be as bad or good as the next.
BUT I know my system is 100% stable, I can run anything at the settings I quote, for extended periods of time. I can bench well into 3Ghz, but I just can't get it stable, as said I might gain from a re-seat.
CCT in torture test is awsome, its soo stressful that the slightest querk will show up. I like prime95, but its outdated now IMO.
GO does bring prime back into the game, but hardly anyone uses it.
So although my gains don't look great to you, its still probably 100mhz more then with it on for my setup.
 
ricco said:
Well you say that, but you 2 might have slightly better CPU's then me to start with. Rem not every IHS will be as bad or good as the next.
BUT I know my system is 100% stable, I can run anything at the settings I quote, for extended periods of time. I can bench well into 3Ghz, but I just can't get it stable, as said I might gain from a re-seat.
CCT in torture test is awsome, its soo stressful that the slightest querk will show up. I like prime95, but its outdated now IMO.
GO does bring prime back into the game, but hardly anyone uses it.
So although my gains don't look great to you, its still probably 100mhz more then with it on for my setup.
Click to expand...


sorry for how i came across. :o
i hear you buddy,point taken.got any links to a guide for the removal?Would you say its tricky?Quite into mods, might give this a go if its explained fully.
And btw your clocks are by no means shabby anyhow.
 
Just so you know, I did'nt read it as hostile - so no appology needed realy.
I am all for deating things, but I would'nt pop the cap off a £200+ CPU for no reason if it was not worth it IMHO.

I can walk you through it and maybe host some pics if you feel the need, but basicly its as easy as stanleyt knife and slowly work in at each edge, not too far though, say 5-6mm, then slowly slide the blade accross to join each corner up. 3sides and it will fold off.
3200wini.jpg
 
ricco said:
Just so you know, I did'nt read it as hostile - so no appology needed realy.
I am all for deating things, but I would'nt pop the cap off a £200+ CPU for no reason if it was not worth it IMHO.

I can walk you through it and maybe host some pics if you feel the need, but basicly its as easy as stanleyt knife and slowly work in at each edge, not too far though, say 5-6mm, then slowly slide the blade accross to join each corner up. 3sides and it will fold off.
3200wini.jpg
Click to expand...

cool mate.looking at the pic it seems really safe, i never realised there were no components anywhere near the edge.what i plan to do is burn this sucker in and find the max stable at 1.6V then to remove the heatspreader and see what difference it achieves on my cpu.100MHz would be very sweet as that should definitely be the 3GHz stable here.

James you should change your sig as you're short-changing yourself. you are not putting 1.525V through it, dont write the BIOS setting in there....your o/c is more impressive than that with the real lower voltage. ;)
 
Concorde Rules said:
Done ;) Gonna have a good game now to test 3D Stability!

Concorde
Click to expand...


thats the stuff.hmm you should be stable at 2.9GHz and 1.5V actual, now if you removed that heatspreader(trapper really) that could be 3GHz at 1.5V. ;)

whqat you playing atm?swat4 is pretty sweet.
 
anybody found any util which lets you change the voltage in windows on ultra-d?
clockgen does nothing and CBI just crashes windows.its a pain in the ass trying to fine tune o/c having to go back to BIOS each time.
 
monkeygabriel said:
anybody found any util which lets you change the voltage in windows on ultra-d?
clockgen does nothing and CBI just crashes windows.its a pain in the ass trying to fine tune o/c having to go back to BIOS each time.
Click to expand...

Clockgen would work if you dropped the start up volts and used the + 123% and over settings.
 
ricco said:
Clockgen would work if you dropped the start up volts and used the + 123% and over settings.
Click to expand...

ahh so set the start up to minimum and choose 123%?then clockgen will let you raise or lower?im talking about raising from 1.45V up, wont 123% be over that already?
 
monkeygabriel said:
ahh so set the start up to minimum and choose 123%?then clockgen will let you raise or lower?im talking about raising from 1.45V up, wont 123% be over that already?
Click to expand...

Yeah, clockgen will only let you use 1.45v max (this is STARTUP setting, not actual), so use a startup of say 1.2v and the +123%(or whatever value you want).

I have tried this and it works.
 
You must log in or register to reply here.
Back
Top Bottom