ati and nvidia speed changes before release??

[jimmyjames123]

The 50 series drivers allowed you to change core and memory frequency? I have a Ti4200, and have not even touched the 50 series "FX" drivers. Didn't want to risk it

 

[AlphaWolf]

The 50 series drivers allowed you to change core and memory frequency? I have a Ti4200, and have not even touched the 50 series "FX" drivers. Didn't want to risk it

Hardocp has a screenshot of it in the 60.xx drivers.

 

[Bad_Boy]

wow...

Albatron said that their 'Geforce6800UV' has core clock running at 600Mhz and the ram is reaching 1Ghz

http://www.pcunleash.com/bbs/zboard...elect_arrange=headnum&desc=asc&no=228
sorry bout the long link. i just found that pretty interesting.

 

[Sandman]

wow...

http://www.pcunleash.com/bbs/zboard...elect_arrange=headnum&desc=asc&no=228
sorry bout the long link. i just found that pretty interesting.

Check out the Aopen card, it only has 1 molex It's also the only one that deviates from the reference design. The pic looks legit too (as legit as any other on that page). Very interesting.



btw, long time lurker, finally registered to point that out

 

[Diplo]

You mean the one that's been in the drivers since the fx5800?

It was around in the Detonators when I had a GeForce 256 (you just needed to use the 'coolbits' registry tweak).

 

[DarN]

Yes, I remember. They took out the need for coolbits with the FX5800 though.

 

[jolle]

wow...

http://www.pcunleash.com/bbs/zboard...elect_arrange=headnum&desc=asc&no=228
sorry bout the long link. i just found that pretty interesting.

Check out the Aopen card, it only has 1 molex It's also the only one that deviates from the reference design. The pic looks legit too (as legit as any other on that page). Very interesting.



btw, long time lurker, finally registered to point that out

There is a off chance that its acctually a Non Ultra board perhaps..
even tho they pass if off as a ultra when displaying it..
Dunno, but if 2 molexes wasnt sortof important for stability, perhaps
Nvidia wouldnt have bothered with it in the first place..
or maybe you rarely acctually HAVE to use them both, they perhaps
just put 2 there cause there is a risk it CAN get unstable if you rely
on only 1.. hmmmmm, dunno...

 

[Psikotiko]

wow...

http://www.pcunleash.com/bbs/zboard...elect_arrange=headnum&desc=asc&no=228
sorry bout the long link. i just found that pretty interesting.

Check out the Aopen card, it only has 1 molex It's also the only one that deviates from the reference design. The pic looks legit too (as legit as any other on that page). Very interesting.



btw, long time lurker, finally registered to point that out

errrr......Old prototype?
Thats the same board that is in the MaximumPC preview.

 

[Bry]

Found this over at Fururemark..Someone attended a leadteck public display of their 6800 nu card and had some 3dmark scores to back it..the posrt can be read here
http://discuss.futuremark.com/forum...p;view=collapsed&sb=5&o=0&fpart=1

Here is what it says

Test result from the show's demo machine
Athlon64 FX-53
Leadtek K8NW Pro
DDRSDRAM 512mb*2
GeForce6800NU (@ 300/500)

3DMark03 result
10285 @ default 1024 X 768

7410 @ 1024 X 768 w/ 4XAA

10285 is quite a bit lower than the scores we were seeing on the 400 MHz 6800U scores..also notice the core and mem clocks..Maybe this will put the rest the 6800NU being reviewed instead of the 8600U theories

 

[AlphaWolf]

Found this over at Fururemark..Someone attended a leadteck public display of their 6800 nu card and had some 3dmark scores to back it..the posrt can be read here
http://discuss.futuremark.com/forum...p;view=collapsed&sb=5&o=0&fpart=1

Here is what it says

Test result from the show's demo machine
Athlon64 FX-53
Leadtek K8NW Pro
DDRSDRAM 512mb*2
GeForce6800NU (@ 300/500)

3DMark03 result
10285 @ default 1024 X 768

7410 @ 1024 X 768 w/ 4XAA

10285 is quite a bit lower than the scores we were seeing on the 400 MHz 6800U scores..also notice the core and mem clocks..Maybe this will put the rest the 6800NU being reviewed instead of the 8600U theories

Well it is a bit, but better than I would expect for a part with 25% less less core clock and 4 less pipes. I guess maybe the fx53 would help some, but still.

Edit: also odd that the pic in the link shows a card with dual molex connectors. I suppose it may not be the actual tested card.

 

[Bry]

Yeah I noticed that..Wonder if that means the Non Ultras are going to have two molex connectors too?? I do not think very many will be happy with that.

 

[Sandman]

There is a off chance that its acctually a Non Ultra board perhaps..
even tho they pass if off as a ultra when displaying it..
Dunno, but if 2 molexes wasnt sortof important for stability, perhaps
Nvidia wouldnt have bothered with it in the first place..
or maybe you rarely acctually HAVE to use them both, they perhaps
just put 2 there cause there is a risk it CAN get unstable if you rely
on only 1.. hmmmmm, dunno...

Yeah, it's on an Ultra's box, which really hints that it's an ultra. Maybe the redesign of the PCB was just enough to eliminate the need for the 2nd one?

 

[jolle]

Yeah, it's on an Ultra's box, which really hints that it's an ultra. Maybe the redesign of the PCB was just enough to eliminate the need for the 2nd one?

That Futuremark link just above has a card with 2 molexes claiming to be
a Non Ultra.. which might just be a clocked down Ultra to "emulate" the
peformance of the Non Ultra, could be flashed with a NU BIOS to hide
the extra pipelines or something, which would be cool cause then you
can perhaps unlock them on a real Non Ultra..

just as that card with 1 molex might be a Non Ultra in reality, they just
put it next to a Ultra box while displaying it..
maybe they screwed up, or didnt have a Non Ultra version ready at the
time.. dunno.. but like you said, its also possible the fiddled around a bit
and found a way to run it stable on 1 molex..

 

[Geeforcer]

10000+ 3Dmark03 for NU? That seems rather impossible, considering that the card should have 56% fillrate of Ultra and Ultra scores ~12000. Unless 3dmark is CPU limited at default resolution, a 300MHz NU card should be scoring around 7000 based on fillrate drop alone.

 

[Tim Murray]

Unless the NU clockspeeds are wrong and are actually higher than the Ultra.

Uttar, you and I need to talk, by the way.

 

[jolle]

what if its a clocked down Ultra, still running 16 pipes..?
Is fillrate really that much of a factor in 1024x786?
the 4x AA is 7400, and with FSAA fillrate becomes more critical..

EDIT
this is a spicy meatball...
http://www.albatron.com.tw/english/news/news_detail.asp?news_id=77
600Mhz core.. that is a tad higher then 400Mhz on the review samples..

 

[DoS]

10000+ 3Dmark03 for NU? That seems rather impossible, considering that the card should have 56% fillrate of Ultra and Ultra scores ~12000. Unless 3dmark is CPU limited at default resolution, a 300MHz NU card should be scoring around 7000 based on fillrate drop alone.

3dmark gets the final score from the four game tests. Tests 2,3 and 4 are not only fill rate limited but shader limited too. It could be that at default resolution (1024x768) the increased shader throughput is more important than fill rate and therefore the relatively high score. Still if the non-ultra part scores with a similar speced CPU around 12000 i find it kind of difficult the NU part to score 10000 with 4 pipes and 100 mhz less on the core...
Unless of course the Ultra is held back a lot by the lack of memory bandwidth in these tests