[DoS]

Hi all,

Due to some very unfortunate circumstances i have lost 2 gfx cards and 350 euros the past to months . However this allowed me to make a direct comparison between these two cards and i would like to share with you all my thoughts
First of all, i intially bought a Club Radeon ATi 9800 Pro. The card was awesome and the way Ati's FSAA made my games look was simply jaw-dropping. However, the card developed a problem. The fan started making wierd noises and it eventually died after only 3 weeks. Because of a not so friendly RMA policy of the store that i had purchased the card from, i decided to replace the fan with an aftermarket solution (otherwise i had to wait two weeks at least). Anyway something went wrong (me was too hasty) and i had one dead card in my hands (scratched pcb, minus one resistor). However i did manage to get an RMA so when the second card arrived i was more carefull. I did change the cooler again with the Artic Cooling solution (the ATi cooler is a bit loud for my tastes) and the card was working fine. Then, one day out of the blue it started artifacting like mad even in the post screen. I have yet been unable to find the exact cause but i am suspecting that the core might have been crashed (i had removed the metalic shim from the gpu to fit the new cooler). I couldn't get another RMA so i decided to get a new card
This time i figured that although all this were probably a highly unfortunate coincidence i would go with an Nvidia board. I had seen some of the 5900 Ultra boards and the definetely looked more solidly manufactured and had quiter coolers (i am the silent comp type). So i got the Asus 5950 (128 Mb). First thing i did was to oc it. Boy can this card oc.... 490 mhz core/946 mhz mem is not bad at all. Then i installed Forceware 53.03 and on to the games.:
In most of the games i couldn't see any speed difference, except some of the older stuff were the FX card seemed a bit faster (I am a competitive on-line player and avid gamer - I dont' benchmark). IQ wise, there is a huge difference between the way the FSAA looks on the two cards. Screenshots can't tell shit, the ATi AA looks much better, so much that at first i thought that setting AA through the control panel was not working on Nvidia. BUT when it comes to filtering, the Nvidia looks a lot better. Seriously, 8xAF is a lot better than ATis, you notice immediately. To be more specific it's much less angle dependent, and if you use the control panel to force it you fon't get trilinear only in the first mip-map band. I know you are going to jump on me now, but i could really notice the effect of that ATi implementation on many games, BF1942 is one. Get a plane in any map (Bocage - fly over the fields) and u ll see what i am talking about. Also in the same game all the maps with a lot of elevation changes/mountains the ATi angle dependent aniso was doing a poor job. Another thing that i should mention is that the Nvidia card seems to have less quirks (general compatibility) all-around.

IMHO i would still get the ATi AA any time (IQ wise). The only problem is that i like it to play it safe and the problems that i run into with 9800 Pro made me get the FX card. Finally, i am dissapointed in ATi putting that problematic sorry-ass of a cooler onto one of their top cards. Also, i still can't understand why sometimes on a cold boot i would get a corrupted screen (that was beofre changing the cooler). Overall, the ATi is the better card, no doubt about it (and when DX9 comes into play it gets ugly) but ATi has to iron-out some issues in the next product cycle if it wants to stay at the top

[vrecan]

I currently run 6 yes count them 6 ati cards all from different companies. Pretty much the whole range from 9500pro to a 9800pro *no xt's* and all with stock coolers and not once has one of them gone out on me in close to two years. So I doubt that you got two bad cards in a row just doesn't seem very likely to me.

The only NV card of late I have had to work with was a horrid 5200 on a friends machine, trying to get that pos to run anything at a decent framerate can be extremely difficult if you want to run anything above 800x600 .

Sorry to hear about your card problems but I cant help to think that your issues had more to do with system heat then two bad coolers.

[Doomtrooper]

Overall, the ATi is the better card, no doubt about it (and when DX9 comes into play it gets ugly) but ATi has to iron-out some issues in the next product cycle if it wants to stay at the top

Funny Stuff....

Tell us how these demos run:

http://www.beyond3d.com/articles/shadercomp/results/

[DoS]

Originally Posted by vrecan

I currently run 6 yes count them 6 ati cards all from different companies. Pretty much the whole range from 9500pro to a 9800pro *no xt's* and all with stock coolers and not once has one of them gone out on me in close to two years. So I doubt that you got two bad cards in a row just doesn't seem very likely to me.

The only NV card of late I have had to work with was a horrid 5200 on a friends machine, trying to get that pos to run anything at a decent framerate can be extremely difficult if you want to run anything above 800x600 .

Sorry to hear about your card problems but I cant help to think that your issues had more to do with system heat then two bad coolers.

Sorry, your assumption is incorrect. No cooling problems here, in fact my pc is pretty well cooled and low-noise. The fact (unless you want to imply that i am lying) that the first's card fan failed after three weeks was disturbing to say the least. And after having examined the fan i can't say i was suprised - can you say "plastic bearing" (perform a search in the Rage3D forums for 9800 pro cooler prolem and u ll see what i mean).
The second card could have been damaged by the removal of the metalic gpu shim (however it was working fine for about a month afterwards) and/or getting progressively crashed by the artic cooling heatsink and i must admit that i am solely responsible for attempting such a thing in the first place and voiding my guarantee. Still, when i pay 400 euros for a gfx card i expect it to come with a half-decent low-noise cooler and i cannot accept that the 9800 pros falls into that category nor explain the random cold-boot artifacting issues or the COD/BF1942 pains (prior to 3.10).
Overall, as i said in my first post, the ATi is by far the better card, it's much more future proof and has better looking AA hands down but for me there are still issues that make me sceptical. As for DX9, i only hope NVidia's driver team start optimizing every shader by hand when PS 2.0 games get standard...

BTW, i am going to dl those demos and report back my findings.

cheers

[Razor04]

Originally Posted by DoS

Still, when i pay 400 euros for a gfx card i expect it to come with a half-decent low-noise cooler

Sorry...but I just can't see how an ATI card with the stock cooler can be louder than the dustbuster...

[Hellbinder]

BUT when it comes to filtering, the Nvidia looks a lot better. Seriously, 8xAF is a lot better than ATis, you notice immediately. To be more specific it's much less angle dependent, and if you use the control panel to force it you fon't get trilinear only in the first mip-map band. I know you are going to jump on me now, but i could really notice the effect of that ATi implementation on many games, BF1942 is one. Get a plane in any map (Bocage - fly over the fields) and u ll see what i am talking about. Also in the same game all the maps with a lot of elevation changes/mountains the ATi angle dependent aniso was doing a poor job. Another thing that i should mention is that the Nvidia card seems to have less quirks (general compatibility) all-around.

That is just totally innacurate...

First of all there is VISIBLE Mip Banding in several games with Nvidias AF. Secondly they are ALSO doing the TRilinear Filter on one stage trick. Thirdly YOU CANT TURN IT OFF most of the time. Thy also use some other AF tricks on a per application basis.

Then when i recently used and Nvidia card you could CLEARLY see that the Textures had a slightly Grungy look. Further there was a BIG difference in the Sharpness of textures in teh difference in games like SWG. While there is a difference becuase of the Angle issue when you set ATi's af to 16x it more than makes up for it.

I simply can not agree with you on this based on my own experience. Nvidia simply has a slightly "Grungy" look to its base textures and not as crisp of an AF affect imo.

[MuFu]

and they say grunge is dead?!

MuFu.

[pino]

Originally Posted by Hellbinder

That is just totally innacurate...

First of all there is VISIBLE Mip Banding in several games with Nvidias AF. Secondly they are ALSO doing the TRilinear Filter on one stage trick. Thirdly YOU CANT TURN IT OFF most of the time. Thy also use some other AF tricks on a per application basis.

Then when i recently used and Nvidia card you could CLEARLY see that the Textures had a slightly Grungy look. Further there was a BIG difference in the Sharpness of textures in teh difference in games like SWG. While there is a difference becuase of the Angle issue when you set ATi's af to 16x it more than makes up for it.

I simply can not agree with you on this based on my own experience. Nvidia simply has a slightly "Grungy" look to its base textures and not as crisp of an AF affect imo.

i disagree, owning a 9800pro in my main pc and 5900nu in my secondary, nvidia AF looks better than ATI's IMHO, maybe it's just subjective, maybe not, i don't know but still i prefer Nvidia aniso.

Anyway 9800 has way better AA and it's faster.

oh btw what is 'grungy' look ?

[gkar1]

LOL this thread is funny. You forgot to mention how dissapointed you were when ATI was caught cheating in quake3

Waste of bandwith is what it is.

[see colon]

LOL this thread is funny. You forgot to mention how dissapointed you were when ATI was caught cheating in quake3

Waste of bandwith is what it is.

why is his opinion about these 2 cards a waste of bandwidth? dos' findings about how nVidia's current offerings compare to ati's mirror that of other people (myself included) who have actualy used both offerings, as opposed to just read reviews.
c:

[Doomtrooper]

Well I wouldn't want you doing reviews then, I tested a 5600 with Brilinear filtering enabled and the mip map boundaries are evident In UT2003 and MAFIA...only a few I tested.
Screenshots are useless but Mike Chambers from Nvnews did expose them in the videos he captured.
To add to that, the brutal performance of the card in question....I also tested about 4 different driver revisions including 53.30 Forceware drivers with Rivatuner and sure enough almost every driver revision is running close to 0.0 LOD BIAS . Setting it -1.0 cleans things up where they should be and a considerable performance hit...i.e not good for graph wars but good for the kids and braggin rights.

[jvd]

DoS first off don't complain about ati cards when you never bought an ati card.


Second coming from owning both a 5800xt and a 9700pro (my sister took the all in wonder 9800pro ) the ati has the best image quality in both af and aa . Hands down.

[demalion]

What's the default LOD bias for 53.03 in "old games"? Maybe OpenGL games like Quake 3 and Serious Sam?
Does UT2k3 and its default LOD bias settings perhaps qualify as an "old" game for this comparison?

[cthellis42]

Originally Posted by gkar1

LOL this thread is funny. You forgot to mention how dissapointed you were when ATI was caught cheating in quake3

Waste of bandwith is what it is.

Yes. They also got caught optimizing for 3DMark03. Good thing we can actually see the scale of the problems, note that they corrected the matters quickly with without issue and in 3DMark03 with both public explaination and apology, and no be hit but year-long FUD-storms, eh? :P

[DoS]

Originally Posted by Hellbinder

That is just totally innacurate...
First of all there is VISIBLE Mip Banding in several games with Nvidias AF. Secondly they are ALSO doing the TRilinear Filter on one stage trick. Thirdly YOU CANT TURN IT OFF most of the time. Thy also use some other AF tricks on a per application basis.
Then when i recently used and Nvidia card you could CLEARLY see that the Textures had a slightly Grungy look. Further there was a BIG difference in the Sharpness of textures in teh difference in games like SWG. While there is a difference becuase of the Angle issue when you set ATi's af to 16x it more than makes up for it.
I simply can not agree with you on this based on my own experience. Nvidia simply has a slightly "Grungy" look to its base textures and not as crisp of an AF affect imo.

Sorry i have to disagree. Forcing 8x AF using the control panel with Catalyst 3.10 produces highly visible mip-banding whereas on the FX5900 with Forceware 53.03 i can't notice any. However i haven't tried UT2003, and even if i did i can't compare it to ATi cause i hadn't tried it with the ATi in the first place and my 9800 pro is no more. Also, using the ATi card with Radlinker to force aniso and leave application-pref filtering (full trilinear in most games) resulted in a highly noticeable performance hit while gaming at 1024x768 4xAA 8xAF in most newer games. I don't know where you came up with the idea that Nvidia only does trilinear in the first mip-map stage, this is something that i hear for the first time and i haven't noticed anything in my games/tests that could imply something like that is happening - show me some proof please
As for the whole brilinear situation i don't know, but if what i am currently experiencing in my games is that, then i do prefer it from ATi's first stage Trilinear, then bilinear control panel implementation. IMO ATi should asap put an extra option in the control panel to allow full trilinear with aniso. In fact, when i first got my 9800 pro i was shocked by how well the IQ was in all of my games, but i kept noticing the annoying mip-bands and couldn't offer an explanation only to find out later and i was really annoyed.

[DoS]

Originally Posted by see colon

why is his opinion about these 2 cards a waste of bandwidth? dos' findings about how nVidia's current offerings compare to ati's mirror that of other people (myself included) who have actualy used both offerings, as opposed to just read reviews.
c:

I learned about this site when i first got my 9800 pro, and found very usefull information in it (the site). Unfortunately the forums seem to have the normal percentage of fanboys as any other forums, although at first i thought differently.

[DoS]

Originally Posted by Doomtrooper

Well I wouldn't want you doing reviews then, I tested a 5600 with Brilinear filtering enabled and the mip map boundaries are evident In UT2003 and MAFIA...only a few I tested.
Screenshots are useless but Mike Chambers from Nvnews did expose them in the videos he captured.
To add to that, the brutal performance of the card in question....I also tested about 4 different driver revisions including 53.30 Forceware drivers with Rivatuner and sure enough almost every driver revision is running close to 0.0 LOD BIAS . Setting it -1.0 cleans things up where they should be and a considerable performance hit...i.e not good for graph wars but good for the kids and braggin rights.

There are some very good articles in IQ and particular related to filtering. I forgot the link, but will search and report back (no not the Anandtech article).

[DoS]

Originally Posted by jvd

DoS first off don't complain about ati cards when you never bought an ati card.
Second coming from owning both a 5800xt and a 9700pro (my sister took the all in wonder 9800pro ) the ati has the best image quality in both af and aa . Hands down.

What the hell is wrong with you ? I said i paid 400 euros (418 including VAT to pe precise) for a Club ATi 9800 Pro back in October. The fan failed after 2 weeks and then all hell broke loose. I ended up with no card for about 2 weeks and then the replacement also broke (after i had it modified cuase i couldn't stand the high pitch noisy fan - the failure could have been caused by my mod although it happened quite a while afterwards). So i thought that ATi is bad luck for me and purchased an Nvidia card which i don't find as bad as every1 thinks here they are. So i should just shut up because you don't agree with me ? I trully regret that i came here in the frist place trying to express my sincere opinion about the two cards. Alright everybody, start blowing ATi's (and each others) horn, then i can go at nvnews or wherever and do the same with Nvidia fanboys and we are all happy.

pffff

[Daliden]

Originally Posted by DoS

What the hell is wrong with you ? I said i paid 400 euros (418 including VAT to pe precise) for a Club ATi 9800 Pro back in October.

Don't mind folks with poor reading comprehension

Image quality is a funny thing, by the way; people pay attention to different things. Let's take AA for example -- I've so far managed to "not see" aliased edges. So AA quality would not be a high priority for me. But for others, it would be the first thing they judge. I'm pretty positive the same goes for the different kinds of texture banding.

I think this board does have its share of ATi fanatics, and that sometimes seems to deteriorate certain threads (it seems fashionable to bash NV for *everything*), but the neutral contingent isn't small either, which is good.

[Pete]

Thanks for taking the time to post your experience, DoS. Don't mind some of the angrier replies that seem to believe that conformity is godliness.

jvd might be thinking that because you didn't buy an ATi ATi card () that you can't generalize to all ATi-based cards. By his logic, no one's bought an nVidia card, either, so all the reviews we've read are hooey. Seriously, his point (it's in there somewhere, you just have to dig for it) that generalizing from a small number of personal experiences isn't wise is well taken. You might have had better luck with that HIS Excalibur Radeon with the ICE (or whatever) cooler.

Hmm, I'm sensitive to both jaggies and MIP-maps, so it's a tough trade-off between ATi's superior AA and nV's superior AF. Tell me, in your experience, does high-res (12x9 or 16x12) on a 19" monitor make AA less important?

[DoS]

Originally Posted by Pete

Thanks for taking the time to post your experience, DoS. Don't mind some of the angrier replies that seem to believe that conformity is godliness.

jvd might be thinking that because you didn't buy an ATi ATi card () that you can't generalize to all ATi-based cards. By his logic, no one's bought an nVidia card, either, so all the reviews we've read are hooey. Seriously, his point (it's in there somewhere, you just have to dig for it) that generalizing from a small number of personal experiences isn't wise is well taken. You might have had better luck with that HIS Excalibur Radeon with the ICE (or whatever) cooler.

Hmm, I'm sensitive to both jaggies and MIP-maps, so it's a tough trade-off between ATi's superior AA and nV's superior AF. Tell me, in your experience, does high-res (12x9 or 16x12) on a 19" monitor make AA less important?

You are spot on when you say that we shouldn't generalise from personal experiences and i realise i may somehow have (a little bit), but i really didn't want to (however the Club Radeon 3D is actually built by ATi). Anyway, i still believe the ATi is the better card overall, and if i could find good brand cards with custom coolers "reasonably priced" were i live (which is not the case, i can only find very expensive Asus XTs and Hercules Pros that are almost as expensive as the Asus XTs) the only thing that would make me think (and that would be just for a moment) in going yet again for another 9800 are a few minor compatibility issues,which are getting ironed-out by the excellent catalyst driver team (which has gone a long way).
As for your IQ question, i happen to game on a 19" Iiyama Vision Master pro 454 (pretty nice monitor IMO) and i too am very sensitive in alliasing and mib-bands but i usually game at 1024x768 4xAA 8xAF cause i am also very sensitivet to average fps
But consider that, with ATi at 1024x768 jaggies were not an issue, never not in a single game. With the FX5900 (and 4xAA again) they are, so much that i am seriously considering trading fps for higher resolutions in many games. Also one more thing that i forgot to mention is that the ATi card definetely looked better in 2D quality and the colors were...well to put it simply"better" than the FX card. Still, all that nview, dualhead feautres in Forceware are pretty cool.

[Pete]

Would you prefer 10x7 w/AA+AF on a 9800 over 12x10 or 16x12 with just AF on a 5900? I'm wondering if I'd prefer nV's AF more and if I could get around the need for AA with a high res on my 19" Mitsu. It can hit 16x12, but sadly only at 70Hz, so 12x9/10 is my realistic max (maybe I could hit 14x10 in certain games at 75+Hz).

[Doomtrooper]

Originally Posted by DoS

BTW, i am going to dl those demos and report back my findings.

cheers

What are the results from the B3D shader competition demos

[DoS]

Originally Posted by Doomtrooper

What are the results from the B3D shader competition demos

I managed to run only the 2nd Place Translucency demo, which looked nice and performed withn the range of 300 ~ 750 fps (mainly depending on the zoom). All the others were either not doing anything at all or prompting "No compatible device found" error messages.

[DoS]

Originally Posted by Pete

Would you prefer 10x7 w/AA+AF on a 9800 over 12x10 or 16x12 with just AF on a 5900? I'm wondering if I'd prefer nV's AF more and if I could get around the need for AA with a high res on my 19" Mitsu. It can hit 16x12, but sadly only at 70Hz, so 12x9/10 is my realistic max (maybe I could hit 14x10 in certain games at 75+Hz).

i haven't tried that yet. Actually i haven't tried any res above 1024 without AA, however i will when i get back home (and considering that my monitor supports 1600x1200 @ 85 Hz i should have tried it a long time ago)