First REAL FX Benchmarks

[MuFu]

LOL.

 

[LeStoffer]

Hmmm, these scores are indeed less than stellar... Maybe they could be off, but if they're not, I wonder how nVidia is going to try and put the GeForce FX in a good light with the reviewers.

 

[Evildeus]

Hmmm, these scores are indeed less than stellar... Maybe they could be off, but if they're not, I wonder how nVidia is going to try and put the GeForce FX in a good light with the reviewers.

How, well hmmm... it's not possible

 

[demalion]

On a 2.53 GHz P4, Wavey was getting about 99 fps for the 9700 Pro with the ExtremeGFX add-on (with no aniso) for SS:SE, using the "The Citadel" demo for the benchmark.

Same specs on the machine, driver changes got the 9700 Pro up to 107. The 9500 Pro started to fall away more (down to 84) indicating bandwidth being an issue, yet the 9500 non pro falls away more (down to 60).

They are reporting 94.6 for the 9700 Pro on a P4 3.066 GHz. With a much lesser bandwidth disparity and higher fillrate relative to the 9700 Pro, I can easily buy the GF FX leading, but as to the degree (which is lower than I'd expect with no AA on at that resolution)...

I'm not going to dig through their poor excuse for a benchmark "article" (strikes me more of a gallery...of both benchmark images and advertisements EDIT: excuse me, the ads are only for the site itself, it seems the gallery style is to "encourage" you to pay them to get the info in a more useful form instead) to find details...do they specify what settings they are using and what demo they are using for the benchmark?

As it stands I don't know that we can have much confidence in what we've "learned".

 

[Doomtrooper]

Link is down now...doesn't take you anywhere. We still have the hyperlinked images though.

 

[Mummy]

Hmm.. i wonder what will be the fate of FX-5800 (not ultra), 400/800.. looks like it'll loose quite much compared to the 9700 non pro, hummm...

 

[antlers]

Reading nvnews.net, looks like the NDA wasn't supposed to expire until tomorrow at noon pacific, so they've pulled the benches down.

 

[Sabastian]

Reading nvnews.net, looks like the NDA wasn't supposed to expire until tomorrow at noon pacific, so they've pulled the benches down.

Hrm, I just got done looking at the review. I think possibly they were experiencing a lot of "hits" as it appears to be up still.

 

[Hellbinder]

Well his scores make no sense.. in that case (meaning the new overclocked results).

SORRY, but i have postet wrong benchmark-results!


these are the right score!!!!!!!!!

Radeon 9700PRO @ ...

398/754MHz: 130fps
425/783MHz: 137,4 fps

vs. GFFX 500/1000 143 fps



i am sorry


*UPDATE No.2*
... but i lost only 9% performance with 4xFSAA!! (118fps)

Because on my system at 375/360 I get 132 FPS, and thats with it running with some wierd quirks. I am pretty confident that the R350 wont have those issues.

 

[Bambers]

Hrm, I just got done looking at the review. I think possibly they were experiencing a lot of "hits" as it appears to be up still.

Sites still working fine for me but if you go to the main page the link to the gf fx benchmarks points at the homepage.

 

[Sabastian]

Hrm, I just got done looking at the review. I think possibly they were experiencing a lot of "hits" as it appears to be up still.

Sites still working fine for me but if you go to the main page the link to the gf fx benchmarks points at the homepage.

Try this linky.

http://www.tecchannel.de/hardware/1109/

 

[Tagrineth]

Botmatch: Err, what the heck? I'm sorry but the 2X FSAA scores are simply impossible. The Ti4600 cannot be that fast with 2X FSAA & 4:1 AF, it does more like 45FPS AFAIK... It simply gotta be a typo. The FX Ultra & 9700 Pro being on par there, however, seems possible.

Look at the 9700 Pro scores, then look at the Ti4600 scores, then look at the 9700 Pro scores again.

 

[nelg]

I also wonder how long you can strain this card before the thermal protection option in the driver kicks in ? If this GPU heats up high and fast and therfore has to down clock all these bechmarks would be considered pruley academic.

 

[Bambers]

Try this linky.

http://www.tecchannel.de/hardware/1109/

Dont think i made myself clear there

I meant the review still works fine for me but you need to know the link like the one you posted

If you go to the homepage the review link there just goes straight back to the homepage unless theres another one somewhere but I can't read german. :|

 

[T2k]

Uttar, start praying!

Err, what for? Sorry, I just didn't understand
Uttar

GFFX will be beated by the 9700 - it seems to me obvious...

Ed: typo

 

[Sabastian]

Try this linky.

http://www.tecchannel.de/hardware/1109/

Dont think i made myself clear there

I meant the review still works fine for me but you need to know the link like the one you posted

If you go to the homepage the review link there just goes straight back to the homepage unless theres another one somewhere but I can't read german. :|

Yeah I realised that when I went to the home page. I did get it BTW.

Here is the conclusion translated with babelfish.

Result

Expectations at the GeForce FX were high, but the reality sobering with the GeForce FX 5800 Ultra disappoints NVIDIA. The chip in the overall evaluation can strike the ATI RADEON 9700 pro - however only scarcely. In some applications the NVIDIA chip lies even in the back.

Weak achievement: Despite high Vorschusslorbeeren the GeForce 5800 Ultra can push itself only scarcely in the total valuation before the RADEON 9700 pro.

The GeForce FX has many technical innovations, which can accelerate the performance of 3D-Anwendungen. But like already in the past those must implement diagram programming this only into the applications.

Whether a further optimization of the drivers brings still another significant performance thrust, remains being waiting.

For the small performance gain one must seize deeply into the bag. First diagram maps with the NVIDIA GeForce FX 5800 Ultra are to cost about 650 euro. One already receives ATI Konkurrenzprodukte with the RADEON 9700 pro for 400 euro.

Detailed information to the features and the chip architecture of the NVIDIA GeForce FX finds you under GeForceFX: The new generation."

Those benchmarks are surprisingly weak given what was expected, especially from an Ultra version of the FX. We obviously need to have confirmation from other reliable hardware sites and further info (about drivers, etc) before reaching any conclusions. If this turns out to be even close to the best performance the high-end FX can offer at that price tag, there will be a major shakeup in the desktop OEM market in 2003.

 

[MuFu]

Despite high Vorschusslorbeeren...

Very disappointing considering the high Vorschusslorbeeren, I agree.

MuFu.

 

[Dave Baumann]

No, I meant the non-Pro... cheapest price I have seen is about £185. I thought choosing that instead of that of the 9700 Pro (which I haven't seen for £220, lol no) would place the GFFX prices in context a little better (especially the non-Ultra) - a card with 90-100% of the performance for less than half the price? Not even considering the 9500NP-to-9700 mod or fact that you can clock a 9700NP to 9700P spec pretty easily.

Are you looking at Komplett? They were advertyising them at about £160 ish the other day - they appear to have gone up.

Anyway, I'm currently doing a 9700 non-pro (the only R300 we haven't reviewed yet) and will be comparing it agains 9500 PRO and 9700 PRO, so you might want to wait for that to make up your mind.

 

[Slides]

Very disappointing considering the high Vorschusslorbeeren, I agree.

MuFu.

Yeah, that high Vorschusslorbeeren always gets the better of Nvidia.

 

[Dave Baumann]

I also wonder how long you can strain this card before the thermal protection option in the driver kicks in ? If this GPU heats up high and fast and therfore has to down clock all these bechmarks would be considered pruley academic.

A 'stress' test might be interesting. Run 3DMark, get the results, leave the demo looping for a few hours and then run the benchmark again.

BTW - can anyone how to measure the power consumption?