NVIDIA: Beyond G80...

[INKster]

How about 1.5GB of RAM for a single GPU ?
Nvidia has just announced the Quadro FX 4600 and 5600, besides a new Quadro Plex model IV:
http://www.nvidia.com/page/qfx_uhe.html

The QFX 5600 PCB and the cooler are longer than the 8800 GTX, i think.

Another interesting fact:
The PCB of the QuadroFX 4600 looks identical to the one on 8800 GTS, but the specs on NV's site state a 384bit bus and 768MB...

 

[trinibwoy]

Hmmm according to Nvidia's numbers these new parts don't seem to be benching much higher than the G7x based Quadros.

 

[KimB]

those digit life results are very suspect, some of the gains look way too big.

http://www.ixbt.com/video3/g80-6-d.shtml#p16

look at the 2560x1600 with aa af. 43 to 63 fps???????

Gains like this can occur if there is a bug in the silicon that prevents full operation at the highest resolutions. This sort of thing has happened in the past.

 

[trinibwoy]

Some anecdotal evidence of shader clocks making a big difference - http://www.overclock3d.net/reviews....0gts__xxx__vs_gainward_8800gts__golden_sample_

G80 seems quite balanced overall though.

 

[icecold1983]

Gains like this can occur if there is a bug in the silicon that prevents full operation at the highest resolutions. This sort of thing has happened in the past.

how would overclocking fix it tho? would this require a respin to fix?

 

[KimB]

how would overclocking fix it tho? would this require a respin to fix?

Yes, it would be the respin, not the higher clockspeed, that would fix the issue.

 

[icecold1983]

if thats the case, why would a modest overclock increase the performance by 40%?

 

[KimB]

if thats the case, why would a modest overclock increase the performance by 40%?

Uh, did it?

 

[icecold1983]

Uh, did it?

yes, performance increased from 43 fps to 63 fps at 2560x1600

 

[KimB]

yes, performance increased from 43 fps to 63 fps at 2560x1600

Where is this comparison?

I seriously don't want you to just spout random information at me without sourcing it.

 

[icecold1983]

uhh i already put the link when i originally brought this up, here it is again

http://www.ixbt.com/video3/g80-6-d.shtml#p16

theres also other games that show abnormally large increases

 

[Pete]

Where is this comparison?

You can find a link to the Digit-Life summary of those iXBT #s icecold linked in Kocur's post further up the page. There are quite a few disproportionally large gains relative to the 8-11% overclocks.

 

[icecold1983]

chalnoth can u oc ur 8800 and give us some comparisons benches? im guessing its a user error on digitlifes part

 

[KimB]

That seems odd. But it is worth noticing that that's not just an overclock they're comparing, but two different brands. Usually this isn't an issue, but it's possible. Anyway, I suppose I could test it myself. Shouldn't take too terribly long. But I'm a bit tired so maybe after I take a nap.

 

[psurge]

Could it be that the non-overclocked results are from an earlier 8800 review (that is, using older drivers)?

 

[KimB]

Could it be that the non-overclocked results are from an earlier 8800 review (that is, using older drivers)?

That would make a lot of sense. Or it could be a reference board vs. retail and some issue or other was fixed.

 

[Pete]

It's possible they substituted a reference board's results for the Palit they were supposedly comparing to the eVGA.* In that case, the reference board may well have used older drivers, but D-L indicates they used 97.92 for both 8800s.

-----
* I could interpret this review quote in that way: "Performance of the standard 8800 GTX card has been reviewed many times. The card from Palit is no different."

 

[KimB]

Just fyi, I checked on my system and it won't overclock stably. I could try underclocking it, I suppose.

Edit: Odd, checked some more, and it's got to be a problem with nTune. I was going by the built-in overclocking test, which fails from just a modest overclock. But I checked back, and it fails from any overclock. So I guess I'll have to be not lazy and just test it manually.

Edit 2: Well, I haven't been able to overclock the thing. nTune refuses to overclock it when the built in test fails, and it fails for every setting outside the default. PowerStrip doesn't seem to be able to overclock it. I'm downloading the 101.41 Vista drivers from guru3D at the moment, so maybe that'll work. We'll see.

 

[Razor1]

the best way to overclock stabily is to actually flash the bios with Nibitor

 

[CarstenS]

If you're not running Vista, my best guess is – weirdly – Ati Tool 0.27 beta. Clocked my GTX according to plan, verified with Rivatuners monitoring.