9700 is great, shouldn't it be better?

[Matthyahuw]

These benchies are confusing me...
Please don't go flaming now, I love what this card can do, but something's up...
http://www.anandtech.com/video/showdoc.html?i=1683&p=9
http://www.anandtech.com/video/showdoc.html?i=1683&p=9
The 9700 only beats the GF4 in Comanche4 by 3fps @ 1600!
And JK2 by 15fps @ 1600...
the GF4 beat the 9700 on both @ 1024 as well...
Did I miss something?

[netviper13]

I'm not sure of the explanation, but I do know that the gap widens considerably when AA and Aniso are used.

[macro6]

its because the benchmark is CPU limited.


look here: http://www.gamersdepot.com/hardware/video_cards/ati/9700pro/005.htm

[StealthHawk]

yes, Jedi Knight is very CPU dependant..Commanche is as well, although nvidia cards always seem to win in it anyway.

[-=DVS=-]

Originally posted by Matthyahuw
These benchies are confusing me...
Please don't go flaming now, I love what this card can do, but something's up...
http://www.anandtech.com/video/showdoc.html?i=1683&p=9
http://www.anandtech.com/video/showdoc.html?i=1683&p=9
The 9700 only beats the GF4 in Comanche4 by 3fps @ 1600!
And JK2 by 15fps @ 1600...
the GF4 beat the 9700 on both @ 1024 as well...
Did I miss something?

Well you missed alot :D

Notice how performance does not change at all on Radeon 9700 its CPU or Driver limited JK2 http://firingsquad.gamers.com/hardware/r300/page21.asp
Shees you missed Stelar AA performance http://firingsquad.gamers.com/hardware/r300/page25.asp:eek:
Or Aniso and AA enabled :eek:
http://firingsquad.gamers.com/hardware/r300/page28.asp

Better yet Check this out best showoff :eek:

http://www.hothardware.com/hh_files/S&V/radeon9700pro(6).shtml


Now how can anyone say GF4 is better lol :D

[JohnsonLKD]

Ill just wait until something makes more than 100 fps with codecreature mark.

MMM...

Yeah. Final answer!

[-=DVS=-]

Some people complain about Commanche 4 Performance :rolleyes:

http://www.hothardware.com/reviews/images/r9700/comanch1280aa.png

http://www.hothardware.com/hh_files/S&V/radeon9700pro(7).shtml :eek:

[SurfMonkey]

Although the the R300 is insanely fast, as the GF4 when it was released, I'd find it hard to justify spending almost twice as much on a new card for less than twice the performance gain.

Since nothing is going to make use of the newer features of the card, at least until DX9 is released, it just seems to make sense to wait. I get all the eye candy I need in the few games I play (NWN, DS etc).

I think January will be good time to look at upgrading. There will be quite a few more options then, Trident, nVidia, and ATi, and DX9 will have been released. Then we'll have some nice demos and maybe someone will finally get to review the other features of the cards that are also interesting such as DVD playback quality and offline 3D modelling.

[Joe Shmoe]

Originally posted by SurfMonkey
Although the the R300 is insanely fast, as the GF4 when it was released, I'd find it hard to justify spending almost twice as much on a new card for less than twice the performance gain.

The 9700 seems to be twice as fast (or more) when you enable AA and anisotropic filtering. Why wouldn't you want that?

[StealthHawk]

Originally posted by Joe Shmoe

The 9700 seems to be twice as fast (or more) when you enable AA and anisotropic filtering. Why wouldn't you want that?

i want 8x FSAA or some method that gets rid of jaggies! at B3D i saw little to no difference in 4x -> 6x. the card shoudl be able to handle 8x with decent speed

[Jazz]

Originally posted by StealthHawk


i want 8x FSAA or some method that gets rid of jaggies! at B3D i saw little to no difference in 4x -> 6x. the card shoudl be able to handle 8x with decent speed

i don't think even the nv30 will be able to handle 8x FSAA with good speed

look at the pics at the bottom on this (http://www.beyond3d.com/reviews/ati/radeon9700pro/index.php?page=page14.inc#quality) page and ull see that 4x removes most of the visible "jaggies" (hehe nice word) so the difference between 4x and 6x is little as most of the edges r quite smooth. Hothardware even said that 6x FSAA is better than Matroxs 16x FAA so i think it must be real good.

with the r400 and nv30 ill think we could see 8x FSAA be used and still be playable :)

[Fotis]

Originally posted by Matthyahuw
These benchies are confusing me...
Please don't go flaming now, I love what this card can do, but something's up...
http://www.anandtech.com/video/showdoc.html?i=1683&p=9
http://www.anandtech.com/video/showdoc.html?i=1683&p=9
The 9700 only beats the GF4 in Comanche4 by 3fps @ 1600!
And JK2 by 15fps @ 1600...
the GF4 beat the 9700 on both @ 1024 as well...
Did I miss something? \
Results with radeon 9700 4x fsaa&16xaniso-gf4 4xfsaa& only 4xaniso.
UT2003 1024x768 antalus Radeon=66-Gf4=30.Well I guess 100% is just slow.Lets move on.
UT2003 1600x1200 antalus Radeon=31-Gf4=11.Hmm 200% faster?No no no that's still a bit slow.
Jedi Knight 2 1280x1024 Radeon=139-Gf4=50.About 165% faster.
Jedi Knight 2 1600x1200 Radeon=101-Gf4=21.It's about 400% faster,not even 1000%,well lets hope for better drives!!
The nice part is that anand didn't use same setting aniso for Gf4 for the simple reason that it would get A VERY LARGE ZERO!!!
No need to flame the results speak for themselves.
The number you have posted are of cpu limited games.The -1 or 3 fps will be corrected in a future driver release.
Please don't compare the Radeon 9700 with the Gf4 it's ridiculous and you know it!!:rolleyes:
I'm waiting to see what nv30 brings to the table.

[Cotita]

You shoult take into account that the nvidia drivers are very mature and optimized and the ATI drivers are very early.

Some benchmarks show that the radeon 9700 pro is held back by the drivers. so future releases will increase performance.

[Matthyahuw]

Don't give me benchies with AA, and FSAA...we both know nVidia and ATi do it differently, I want them tested on the SAME grounds...:rolleyes:
I KNOW the 9700 is going to wipe the floor with the GF4s when AA and FSAA is on, especially with ATi's semi-AA filtering method.

[volt]

Originally posted by DVS
Some people complain about Commanche 4 Performance :rolleyes:

http://www.hothardware.com/reviews/images/r9700/comanch1280aa.png

http://www.hothardware.com/hh_files/S&V/radeon9700pro(7).shtml :eek:

And what did they test it on? Cuz I see something totally opposite. I just finished watchin "The Screensavers" on TechTV and their 9700 Pro Review. Check it out at http://www.techtv.com/screensavers/products/story/0,24330,3396472,00.html

Now if you look at Commanche 4 benchmark, you will notice something weird :) Other benchmarks look VERY impressive IMO. It was a good review I think, so check it out.


Comanche 4

1280x1024 32-bit color / 4X AA

Radeon 9700 Pro - 25.76
GeForce4 Ti 4600 128MB - 25.0
Radeon 8500 128MB - 18.64

[jbirney]

Just look at AA benchies then since the both use a MSAA approach with the biggest difference is that R9700 has a better choice for the sampling pattern which usally results in better jaggie removall. But that not the reason. The real reason is that both games are heavily CPU bounded:


Taken from the HardOCP

Comanche shows that at pretty much any resolution with no AA or AF, we are CPU limited.

http://www.hardocp.com/article.html?art=MzQwLDI=

More from that review:

Can you say, "CPU limited"? Yeah, I thought you could. It is however notable that the Ti4600 and the Parhelia are far from CPU limited.

http://www.hardocp.com/article.html?art=MzQwLDQ=



http://www.hardocp.com/images/articles/1029734693GLECcBBUFk_4_5.gif



The above dives home the point of JK being CPU limited on the R9700, no?

My advice is to look at all of the reviews before you make a judgement. For some reason you may find differences between review sights. Look for a common pattern as that is usally correct.

[Jazz]

Originally posted by Matthyahuw
[BI KNOW the 9700 is going to wipe the floor with the GF4s when AA and FSAA is on, especially with ATi's semi-AA filtering method. [/B]

Well their "semi" filtering method is superior to nvidias "complete" filtering method so i don't see what the problem is :confused:

like others have said the r9700 is cpu limited

[John Reynolds]

Originally posted by Matthyahuw
Don't give me benchies with AA, and FSAA...we both know nVidia and ATi do it differently, I want them tested on the SAME grounds...:rolleyes:
I KNOW the 9700 is going to wipe the floor with the GF4s when AA and FSAA is on, especially with ATi's semi-AA filtering method.

Oh, but it was perfectly fair to compare the GF3/GF4 multisampling scores against the 8500's supersampling scores, right?

And I have no idea where you're pulling the semi-AA statement from. Or are you just parroting Kirk's claims about, presumably, the 8500?

[Matthyahuw]

Originally posted by John Reynolds
Oh, but it was perfectly fair to compare the GF3/GF4 multisampling scores against the 8500's supersampling scores, right?What? when did I do this?
I'M NOT BASHING THE RADEON! IT'S A GREAT CARD, CHILL!
I just want an equal comparison...GEEEEEZ
And I have no idea where you're pulling the semi-AA statement from. Or are you just parroting Kirk's claims about, presumably, the 8500? Everyone here knows that ATi doesn't do AA everywhere, I don't know why you all are arguing against this point...

[saturnotaku]

Matt, I don't think JohnReynolds wasn't responding to you, but to comments made by others (in this thread and others). I was about to make his point myself. It was perfectly fair to compare the GF4 Ti to the 8500, yet you can't do so between the GF4 and the 9700? Sounds like typical fanATIc rhetoric to me, unless someone can prove me wrong.

[Bigus Dickus]

Originally posted by Matthyahuw
What? when did I do this?
I'M NOT BASHING THE RADEON! IT'S A GREAT CARD, CHILL!
I just want an equal comparison...GEEEEEZ


Equal comparison... yeah, I'll give you that. The 9700 looks better than the GF4 in almost all of the screenshots I've seen, so it is a bit unfair to the Radeon 9700. Unfortunately, it's simply impossible to adjust AA quality and AF quality on the GF4 to match the 9700, so we'll have to do with these somewhat NVIDIA favoring comparisons.

:rolleyes:

Everyone here knows that ATi doesn't do AA everywhere, I don't know why you all are arguing against this point...
Pass whatever that is that you're smoking. The only thing that the 9700 doesn't Anti-Alias is alpha textures, the same as the GF3/GF4. Although the 9700 supports a feature to address alpha textures in OpenGL (which might be exposed in later drivers) and the GF4 doesn't.

Perhaps you meant AF instead of AA (Anisotropic Filtering instead of Anti-Aliasing)? In that case, you should know that the Z-rotation drawback of the 8500's algorithm has been addressed, and the 9700 now offers improved AF in combination with trilinear filtering, as well as higher sampling rates than the GF4. In short, the 9700's Anisotropic Filtering is both visually superior and faster.

Doesn't do AA everywhere (or AF if that's what you meant).... pfft. :rolleyes:

[Fotis]

There we go again with Ati's broken features!!!:rolleyes:

[John Reynolds]

Originally posted by saturnotaku
Matt, I don't think JohnReynolds wasn't responding to you, but to comments made by others (in this thread and others). I was about to make his point myself. It was perfectly fair to compare the GF4 Ti to the 8500, yet you can't do so between the GF4 and the 9700? Sounds like typical fanATIc rhetoric to me, unless someone can prove me wrong.

Well, I don't see why someone would say it's unfair to compare the AA scores since both cards use multisampling. I can see someone arguing against making comparisons at all in initial previews/reviews, though.

The one thing no reviewer who compared the two products did was to establish an IQ baseline (LOD, texture filtering, edge AA quality, etc.) and work from there.

[jbirney]

It was perfectly fair to compare the GF4 Ti to the 8500, yet you can't do so between the GF4 and the 9700? Sounds like typical fanATIc rhetoric to me, unless someone can prove me wrong.

Actaully IQ wise its was not fair to compaire the 8500 to GF4 only because it was not and never could be an apples to apples comparison. The GF4 had a more robust form of AF but ATIs was faster and pretty good in most cases. The 8500 has SS for its AA method wich provided for SOME texture filtering where as the MS method of the GF4 did not (as well as can not handle the alphas). But comparisons were made as this was what was out at the time.

Now those same things are still valid as this its what is out now. The R9700 has a higer AF IQ setting and use a simular method for AA. But its still not a true apples to apples. However since that will never really happen its not too far off to compare the two.

[Floor]

Originally posted by Joe Shmoe

The 9700 seems to be twice as fast (or more) when you enable AA and anisotropic filtering. Why wouldn't you want that?

Probably has something to do with the fact that smarter people wait to see what the competition releases before diving in head first.