Is the 9800 pro just the 9700 overclocked?

[GamertilIdie]

Title says it all

 

[dguy6789]

No, it has higher clock speeds and some gpu enhancements. Anandtech's test showed, that at the same clocks, the 9800 pro was 30% faster then the 9700 pro in AA and AF. Not much difference otherwise though.

 

[Mingon]

the 9800 pro was 30% faster then the 9700 pro in AA and AF. Not much difference otherwise though.

think you have added a 0 to your 3 by accident

 

[Budmantom]

Originally posted by: Mingon

the 9800 pro was 30% faster then the 9700 pro in AA and AF. Not much difference otherwise though.

think you have added a 0 to your 3 by accident

I don't think it was an accident (in certain applications it can be 30% faster).

 

[BenSkywalker]

Is the 9800 pro just the 9700 overclocked?

Current evidence says yes. The R9800 is using a different code path in the drivers that when enabled on the R9700 pro performs nigh identical to the R9800 clock for clock(as much as two different boards will). It may be that they have done some enhancements to the core, the FBuffer sounds as if it would fall under that category, but as of right now performance indicates that the boards are simply different clocked varriations of the same chip(expect the same thing when the R9600Pro finally shows up, it will appear to be as fast/faster then the R9500Pro because of a code path in the drivers to make it look that way, ATi will likely enable that code path for the original R300 core boards after the initial wave of press is over).

 

[Glitchny]

other than somne AA/AF tweaks and the fbuffer, its basicly jsut an overclocked 9700pro, so its just really damn fast

 

[BenSkywalker]

other than somne AA/AF tweaks

The AA/AF tweaks are working on R9700s with the driver tweak in place(along with the HyperZ functionality).

 

[dguy6789]

Originally posted by: Mingon

the 9800 pro was 30% faster then the 9700 pro in AA and AF. Not much difference otherwise though.

think you have added a 0 to your 3 by accident

Well, sometimes it is indeed 30% faster, others it is just as fast, it varies a large margin, so theorectically, your right as in most cases thats about right.

 

[Batman5177]

what about if we softmod the 9700 pro to a 9800 pro, then clock the modded 9700 pro to 9800 pro speeds?

anyone have the resources to test this out?

 

[touchmyichi]

hard ocp had a interesting article on this. 9700 pro vs 9800 pro

 

[FearoftheNight]

THe softmod comparision has already been done xbitlabs i think. And it shows that ATI is holding back r9700 pro performance on us.

 

[chizow]

Originally posted by: touchmyichi
hard ocp had a interesting article on this. 9700 pro vs 9800 pro

:yawn;

We know that some people are ?hacking? 9800 Pro drivers to work on 9700 Pro?s giving you a small speed boost. This article does not reflect that. We are simply using the exact same driver Catalyst 3.2 for both cards, which detected both for what they were and installed normally, just as many end users would if they were installing the cards on their own system.

They didn't do anything we haven't seen before. 9700pro on the 9700pro path, 9800pro on the 9800pro path.

Digit-Life went and applied the softmod and the results were VERY different.
Don't bother reading the conclusions or blurbs (as I've found out they didn't change the text and conclusions in the article), just look at the graphs and grey boxes with the 9800pro softmod results. You'll see very clearly that the differnces at identical clockspeeds are negligible, and in some cases, favor the 9700pro once softmodded!

I'm not even convinced the F-buffer is a core addition, as transistor counts are identical from the sites that provide it, both ~110 million transistors. Also, from what little information there is on the F-buffer (mostly from Beyond3D), it sounds like the F-buffer is just an allocation of VRAM. ATi provided illustrations are vague at best; the flow-charts don't indicate whether the F-buffer resides in an on-die cache or in VRAM. Easily enabled in drivers, but we probably won't know for sure for some time.

Chiz

 

[SickBeast]

if it is, that's terrible. i'll have to switch to matrox or something

 

[bedrocked]

the 9800pro softmod raised my 9700Pro 3dmark03 by bout 800 points but my 3dmark01SE dropped bout a 1000.

 

[floydsta]

Originally posted by: BenSkywalker

Is the 9800 pro just the 9700 overclocked?

(expect the same thing when the R9600Pro finally shows up, it will appear to be as fast/faster then the R9500Pro because of a code path in the drivers to make it look that way, ATi will likely enable that code path for the original R300 core boards after the initial wave of press is over).

The 9600pro is actually very different to the 9500pro, it is manufactured on a .13 micron process with 4 pipes as opposed to 9500pro on .15 micron process with 8 pipes. The 9600 will have an improved memory controller (better than even the 9800pro due to the extra room from the .13 process and 4 less pipes), some other improved algorithms, and a higher clock, but could well be slower than the 9500pro

 

[BenSkywalker]

The 9600pro is actually very different to the 9500pro

When did I say otherwise? As far as the improved memory controller, ATi is saying the same thing about the R9800 v the R9700 while in fact they are using a driver hack to make it appear that way as of now. I will believe it when I see some results for it. As of right now I haven't seen anything conclusive to show that the RV350 is going to be any different then the R300 on a finer build process. I also have yet to see the R350 show it has any architectural advantage over the R300.

Maybe ATi has built improvements in and its just very poor drivers that are holding it back right now, but with the R300 and "R350" boards performing on an identical basis once you dance around a driver hack it doesn't lead me to believe that that is the case. Of course, I may be giving ATi's driver team WAY too much credit and it could just be they really, really suck