PDA

View Full Version : ATi/Nvidia AF


Pages : [1] 2

Blacklash
11-10-03, 08:57 PM
You see alot of reviews coming out these days placing the mid ranged ati cards against the nvidia. Often times the tests are no AA/AF or 4XAA and 8XAF.

We all know that ati has better and faster AA.

The thing to watch for is this:

http://www.extremetech.com/article2/0,3973,1153876,00.asp

That article explains quite clearly ati AF performs differently in quality and performance mode. When set to 8XAF performance it is like Nvidia's 4XAF, texture sample 32See said article.

If the card is in quality mode you could get a correct comparison, or turn it up to 16XAF in performance which is like the nvidia 8XAF. So don't be duped thinking you are getting a one to one comparison without said factors.

In engrish : Ati performance 16XAF=nvidia 8XAF.

If you are interested in a card you can easily pick one up at best buy or circuit city and test it for a few days and return it if not satisfied.

I am not crazy about Compusa, but the fact remains, at many of them, they will allow you to get the service plan with a card and later trade it in for a better card for the difference.

Both Ati and Nvidia make fine cards, I own both brands and no one would be screwed buying either.

You need to watch the spin doctors in these communites though, it can get quite deep, and the trolls? Ugh. Enough to make on stop browsing the forums all together...

simwiz2
11-10-03, 09:23 PM
Wouldn't most reviews be using the "Quality" setting on both cards?

Blacklash
11-10-03, 09:52 PM
Most people that I know bench in performance mode. If the sites are indeed benching with the ati driver in quality mode then there is no problem.

Again, I find that an odd setting to benchmark performance.

My point is only in that case are you getting a 1to1.

Be aware is what I am saying. If it is not mentioned I would suspect the opposite of what you state. Most old school benchers go with performance driver settings for just that reason, performance.

ragejg
11-10-03, 10:16 PM
I find no sense in benching games in performnce mode... only reason i ever bench in performance mode is to acquire "best possible" 3dmk2k1 and 2k3 scores...

Ok, I think I have the agreement of quite a few ppl here in saying that games only look right in high quality IQ anymore... teh performance settings really make things look ugly (uglier than the old settings in older drivers used to make em)... it seems to me that higher IQ seemed to really take hold when the average enthusiast class system got 12k in 3dmark2k1 @ highest IQ, vsync on, nothing overclocked... that' s when just about EVERYBODY was AT LEAST playing with a lotta AF and some 2XAA in just about everything... once the setups became this p0wahfull, it seemed like nobody really felt the necessity to take IQ shortcuts anymore...

...

hmmm, about the AA thing... (ATI 2X vs. NV 4x) ... maybe someone should come up with a way to compare AA IQ and performance WITHOUT the numbers and names, just with the actual AMOUNT od antialiasing that's taking place... like you could match the AA qualities up and then compare the performance hits... hmmmm....

ChrisRay
11-10-03, 10:20 PM
Originally posted by ragejg
I find no sense in benching games in performnce mode... only reason i ever bench in performance mode is to acquire "best possible" 3dmk2k1 and 2k3 scores...

Ok, I think I have the agreement of quite a few ppl here in saying that games only look right in high quality IQ anymore... teh performance settings really make things look ugly (uglier than the old settings in older drivers used to make em)... it seems to me that higher IQ seemed to really take hold when the average enthusiast class system got 12k in 3dmark2k1 @ highest IQ, vsync on, nothing overclocked... that' s when just about EVERYBODY was AT LEAST playing with a lotta AF and some 2XAA in just about everything... once the setups became this p0wahfull, it seemed like nobody really felt the necessity to take IQ shortcuts anymore...

...

hmmm, about the AA thing... (ATI 2X vs. NV 4x) ... maybe someone should come up with a way to compare AA IQ and performance WITHOUT the numbers and names, just with the actual AMOUNT od antialiasing that's taking place... like you could match the AA qualities up and then compare the performance hits... hmmmm....

You know whats kinda funny about this. Is I use Performance Modes in AF For Both ATI & Nvidia because I just dont see the mip map transistions.

But, One thing I "Do" do. Is Set my Mip Map Quality to High Image Quality On both cards. I cant stand Mip Mapping at low detail.

AnteP
11-10-03, 10:25 PM
I don't think anyone posts review scores using the performance mode, I sure don't and I've never heard or seen anyone else doing it.

When the 5800/9700 were fresh I did try all the different modes though but in the long run you can't keep doing that in each and every review.

Besides viddeocard nowadays shouldn't have to use such simpleton hacks eh? ;)

Blacklash
11-10-03, 10:26 PM
For the hell of it, I would like to see it though. Yes no one plays with mip detail and full performance its muggily, but as you state some use it to inflate scores on 3dmark03.

Just out of raw curiosity I would love to see a site do it. Nvidia and Ati at full performance in driver settings. 4XAA/8AF and 4XAA/16XAF.


EDIT: I fail to see setting BOTH tested cards to performance as a hack. Its in the driver for a reason, to be used. However an unethical spin doctor could easily dupe folks by doing what I describe in post one.

I would rather be aware and ask questions than assume people are being scrupulous, personally.

ChrisRay
11-10-03, 10:56 PM
Originally posted by Malficar
For the hell of it, I would like to see it though. Yes no one plays with mip detail and full performance its muggily, but as you state some use it to inflate scores on 3dmark03.

Just out of raw curiosity I would love to see a site do it. Nvidia and Ati at full performance in driver settings. 4XAA/8AF and 4XAA/16XAF.


EDIT: I fail to see setting BOTH tested cards to performance as a hack. Its in the driver for a reason, to be used. However an unethical spin doctor could easily dupe folks by doing what I describe in post one.

I would rather be aware and ask questions than assume people are being scrupulous, personally.

Performance or High Performance? I guess I'm just wierd. But I seriously dont see most of these differences between Performance And Quality. I prefer performance modes on both cards.

Btw. You realise. Nvidia gives users FAR more control Over Mip Map Detail. Here let me give you an example.

ChrisRay
11-10-03, 11:07 PM
I think it should be noted. You cant reach the highest mip map or the lowest mip map details w/o Using Riva tuner.

But there are some real discrepencies in The difference between ATI and Nvidia @ Highest Performance modes, And Probably Quality too.


Here's High performance Mip Mapping with 8xS AA, High Performance 8x AF, And Lowest Possible Mip Mapping.

ChrisRay
11-10-03, 11:07 PM
Here's 8xS AA, High Performance 8x AF, With Highest Possible Mip Mapping.

ChrisRay
11-10-03, 11:14 PM
And for Comparisons sake. Quality AF, Max Mip Map, 8xS AA

Sazar
11-10-03, 11:21 PM
not much difference in the last 2 caps chris... just some extra sharpness... but the first clip is pretty bad... I dunno if I could even consider gaming with something like that...

you reckon you can post comparitive shots with the ati card... if you still have it ?

Humus
11-10-03, 11:24 PM
Originally posted by Malficar
In engrish : Ati performance 16XAF=nvidia 8XAF.

Not really. The first is 16xAF with bilinear, then other is 8xAF with trilinear (if trilinear is used). That the maximum number of samples add up to the same doesn't make it equal.

Rogozhin
11-10-03, 11:30 PM
Did you notice the max payne billboard in the background ;)

I'll do a shot (I don't have the pro version of the benchmark but I'll see if I can get close with 6xAA and 16xQualityAF on my 9800).

rogo

ChrisRay
11-10-03, 11:36 PM
Originally posted by Rogozhin
Did you notice the max payne billboard in the background ;)

I'll do a shot (I don't have the pro version of the benchmark but I'll see if I can get close with 6xAA and 16xQualityAF on my 9800).

rogo

Get performance too. :) And Dont forget lowest and highest possible Mip Maps. :)


Sazar, I could install the 9500 Pro. But it'd be easier for someone else to do it heh :O

Sazar
11-10-03, 11:45 PM
Originally posted by Rogozhin
Did you notice the max payne billboard in the background ;)

I'll do a shot (I don't have the pro version of the benchmark but I'll see if I can get close with 6xAA and 16xQualityAF on my 9800).

rogo

well it is based on the MP engine init :D

so its expected...

Sazar
11-10-03, 11:46 PM
Originally posted by ChrisRay
Sazar, I could install the 9500 Pro. But it'd be easier for someone else to do it heh :O

yah :)

I can't be arsed to dl 3dmark software.. but nice comparisons...

saturnotaku
11-10-03, 11:55 PM
I pretty much use Performance mode all the time because I honestly can't see that much of a difference between the two and Performance is just that. Unless we're talking about older games (Q3, UT and older) then I'll jump on Quality because it nets me the highest possible IQ. But for something like UT2003 I really can't tell much of a difference between Performance and Quality.


BTW - Sazar, you can make one post when you quote multiple sources. Copy/paste is your friend. It gets annoying having to scroll through your multiple posts when you make a response to each person individually.

ChrisRay
11-11-03, 12:22 AM
Originally posted by saturnotaku
I pretty much use Performance mode all the time because I honestly can't see that much of a difference between the two and Performance is just that. Unless we're talking about older games (Q3, UT and older) then I'll jump on Quality because it nets me the highest possible IQ. But for something like UT2003 I really can't tell much of a difference between Performance and Quality.


BTW - Sazar, you can make one post when you quote multiple sources. Copy/paste is your friend. It gets annoying having to scroll through your multiple posts when you make a response to each person individually.


I agree, I'm Generally always in Performance Modes on My Geforce FX, Because It really does not make a difference to me. I took some still shots comparing 8x High Performance/Performance/Quality and I have to sit at them and stare for minor MINOR things,

However, Going from No AF to 8x High performance is a huge difference.

Sazar
11-11-03, 12:29 AM
Originally posted by saturnotaku
BTW - Sazar, you can make one post when you quote multiple sources. Copy/paste is your friend. It gets annoying having to scroll through your multiple posts when you make a response to each person individually.

I spose you could say the same thing to various other members of this board too saturn unless you think I am the one primarily responsible for this... in which case I would have thought a pm would have been appropriate...

ChrisRay
11-11-03, 12:31 AM
Originally posted by Sazar
I spose you could say the same thing to various other members of this board too saturn unless you think I am the one primarily responsible for this... in which case I would have thought a pm would have been appropriate...

My Fault for the multiple posts :(

Blacklash
11-11-03, 12:39 AM
Morrowind 2xAA/8xAF, Quality settings, 1600x1200.

I don' t think that brilinear filtering is too bad...

Compressed to 1024:

http://members.cox.net/malficar/Morrowind%202003-11-11%2000-03-29-20s.jpg

Large file here:


http://members.cox.net/malficar/Morrowind%202003-11-11%2000-03-29-20.jpg

Rogozhin
11-11-03, 01:09 AM
I can't make the 3dmark screen without a pro version-not close enough to the original to make it valid.

But here is a 4xAA 16xAF quality shot in IL-2 FB with the 1.2beta on a 9800.

http://www.uploadit.org/files2/111103-il2fb-2003-11-02-22-13-10-6.jpg

rogo

Blacklash
11-11-03, 02:14 AM
9800 pro is a sweet card, and as I have said many times the best long term card to purchase.

Nice SS.

I can go 4AA/8AF here but its not going to be in 1600x1200, but rather 1280x960.

Abusing this card til the NV40 rolls out, and if they blow it this time with crappy decision making etc. I likely won't be getting another nvidia card for sometime.

Rogozhin
11-11-03, 02:21 AM
that shot was initally 1280x960 but I reduced it to 1024 for size.

I won't be going nvidia for quite a while (last card I had was the 4600 and their af and aa sucked so bad compared to a measly 8500 that I vowed to not use nvidia in my own rig until they turned over a new leaf-they've **** on that leaf over 3 times now so there is no way).

rogo