Albatron GeForceFX 5200 Ultra Review

[First]

This will be our first review simply because 9800/Pro and 5900/Ultra testing and writing takes a bit longer to complete than budget style add-in card testing. We'll be limiting this review to 1024 and 1280 resolution testing but with AA, AF, etc. benchmarks included of course, and fairly detailed IQ analysis. We'll add in 1600 resolution testing if we have time, but obviously there are very few users that purchase sub $100 cards to run them at 1600x1200 with AA/AF enabled, mostly because budget cards are nearly unplayable at these settings.

Anyway, let me know if there is anything in particular you want to be emphasized in this review given the market this card is aimed at.

Take care,

Evan

 

[Corporate Thug]

planning on including an overclocking section?

 

[cmdrdredd]

I'd focus on image quality at high rez and high AA/AF. This is what most people want. Both cards can do 1024x768 at full AA/AF but which card looks the best doing it?

 

[jjyiz28]

use custom timedemos.

 

[First]

Originally posted by: shady06
planning on including an overclocking section?

Of course.

 

[OneOfTheseDays]

i would like to see more non-standard benchmarks as well, i'm sick of seeing just 3Dmark, quake 3, and unreal tournament.

 

[Schadenfroh]

i would like to see it versus a 4200 and a mx mainly
and dont forsake opengl, to people like me it is just as important (if not more) than direct X

 

[DAPUNISHER]

It would be beneficial to bench it using a mid range CPU, Ram, and board since that's most likely the kind of systems you would find these cards in.

 

[nick1985]

open gl > direct x

 

[jeffrey]

How about testing 1280X960 instead of 1280X1024?
1280X1024 does not fit right on a 4:3 aspect ratio screen.

 

[Mingon]

What about 800x600 with af/aa as that is were the best performance is likely to be - and its the resolution most likely to be used for tv out

 

[First]

Originally posted by: jeffrey
How about testing 1280X960 instead of 1280X1024?
1280X1024 does not fit right on a 4:3 aspect ratio screen.

We've tested 1280x960 in UT2K3 only so far.

Originally posted by: Mingon
What about 800x600 with af/aa as that is were the best performance is likely to be - and its the resolution most likely to be used for tv out

Excellent suggestion, thanks.

I'll also try and dig up a GeForce4 MX440. Otherwise, we'll still have a Ti4200, Ti4600, FX5200, FX5200U, Radeon 9200/9200 Pro and perhaps a mid-range card such as a 9600 or 5600.

 

[thornc]

Originally posted by: Evan Lieb

Originally posted by: jeffrey
I'll also try and dig up a GeForce4 MX440. Otherwise, we'll still have a Ti4200, Ti4600, FX5200, FX5200U, Radeon 9200/9200 Pro and perhaps a mid-range card such as a 9600 or 5600.

This is interesting, I'm plannig on getting a new system soon, and when it comes for the video card I'm completly in doubt about what to choose... a good old Ti4200 retail loaded with goodies or a new 5200(Ultra) ... this review will surely help...

Hope it gets out soon....

 

[jeffrey]

Can you share any info you have regarding the FX5600 Ultra?
This is the one card out of Nvidia's whole FX line that actually looks to have a good price/performance ratio, but it is nowhere to be found.

400mhz core 800mhz memory

 

[lchen66666]

Do you have plan to review the other budget video card, like FX5200(Non Ultra), FX5600(No Ultra)
and 9600(Non Pro) ?

 

[MemberSince97]

I like to see a quick sidenote or speculation of performance compared to New DX9 mobility parts....

 

[Pete]

Actually, I think I saw a PNY 5600U for $250 in one of the Sunday circulars (CompUSA or something).

 

[McArra]

Benchmarks with Unreal II and GTA:vice city would be great

 

[Pete]

Please look into the bilinear/trilinear filtering issue on UT2K3. nV seems to apply bi when you select tri with current drivers (thus boosting scores--direct link for 56k, if it works). It's obvious in still pictures with colored MIP-maps (Beyond3D discussion), but is it detectable in motion?

 

[First]

Originally posted by: Pete
Please look into the bilinear/trilinear filtering issue on UT2K3. nV seems to apply bi when you select tri with current drivers (thus boosting scores--direct link for 56k, if it works). It's obvious in still pictures with colored MIP-maps (Beyond3D discussion), but is it detectable in motion?

As far as I can tell, no. IQ across each segment of NVIDIA and ATi cards is very close in almost all games, including UT2K3.

In any case, I think one of NVIDIA's engineers mentioned to me a while back that trilinear buffering is disabled in certain games (this could apply to certain levels in UT2K3) as it'll significantly degrade fps for no good reason (i.e. no IQ gain). Though I'll have to confirm that info.

 

[Mingon]

So is the review done yet p.s Evan I might have a pentium M coming to try in my 865pe and 875 boards have you tried this yet?

 

[First]

Originally posted by: Mingon
So is the review done yet p.s Evan I might have a pentium M coming to try in my 865pe and 875 boards have you tried this yet?

Yup, benchmarks are completely done at 4 different resolutions at 2X AA/4X AF, 4X AA/8X AF, and no AA or AF. All I have to do is write the review and make sure Anand likes our new format.

And no, I haven't tried a P-M in an 865PE or 875P board. Sounds cool, I know.

 

[Dean]

Originally posted by: Evan Lieb

Originally posted by: Pete
Please look into the bilinear/trilinear filtering issue on UT2K3. nV seems to apply bi when you select tri with current drivers (thus boosting scores--direct link for 56k, if it works). It's obvious in still pictures with colored MIP-maps (Beyond3D discussion), but is it detectable in motion?

As far as I can tell, no. IQ across each segment of NVIDIA and ATi cards is very close in almost all games, including UT2K3.

In any case, I think one of NVIDIA's engineers mentioned to me a while back that trilinear buffering is disabled in certain games (this could apply to certain levels in UT2K3) as it'll significantly degrade fps for no good reason (i.e. no IQ gain). Though I'll have to confirm that info.

Thats the thing! ATI does enable trilinear filtering when running UT. You cannot get an apples to apples comparison when nvidia is disabling it. ATI will always take the performance hit and lose unless you set the ATI drivers to performance. If you run them both at their respective "quality" setting, the ATI card will lose for showing the proper IQ.

 

[Pete]

Originally posted by: Evan Lieb
As far as I can tell, no. IQ across each segment of NVIDIA and ATi cards is very close in almost all games, including UT2K3.

In any case, I think one of NVIDIA's engineers mentioned to me a while back that trilinear buffering is disabled in certain games (this could apply to certain levels in UT2K3) as it'll significantly degrade fps for no good reason (i.e. no IQ gain). Though I'll have to confirm that info.

Interesting, It does fly in the face of nVidia's "Turn it all on!" slogan, though (as eagle-eyed and steel-trap-minded Wavey pointed out). I mean, you can turn it on, but nV will "helpfully" turn it back off for you, but I honestly don't trust nV enough to let them choose my settings for me. Still, I don't have the cards to investigate this further. On the one hand, you say it's unnoticable; OTOH, posters at B3D say it is. Someone lend me a 5900 and 9800P, quick!

 

[DaveBaumann]

Originally posted by: Evan Lieb

In any case, I think one of NVIDIA's engineers mentioned to me a while back that trilinear buffering is disabled in certain games (this could apply to certain levels in UT2K3) as it'll significantly degrade fps for no good reason (i.e. no IQ gain). Though I'll have to confirm that info.

Trilinear Buffering?

But anyway, regarless - you think its OK to just arbitarily remove Trilinear filtering at will? You did not question this when you were told it? Did you actually investigate any of the maps in UT2003 to see if it was noticable or not? Did you solicit opinions from the developers to ask if they thought it was OK to alter the IQ from which they selected for their application?

During the day of the 5900 launch I was at NVIDIA's HQ listening to to all kinds of presentations (which I believe I still have most recorded) which included talks about their new IQ settings in the 44.03 drivers. Not once did they say to us that they were purposefully turning Trilinear off becuase it offers no IQ increase, in fact all I took from that was that the Quality mode offered the same quality as the old Application mode (i.e. Trilinear all the time) but done in a more optimal fashion.

Come on, really, this is 2003 - Trilinear filtering has been here for years (and even for Free on NV10, in terms of fill-rate), should we really be accepting that its just turned off on these new "uber boards" being released which are several generations on from those boards that initially offered it?