Radeon 9700 vs. GF4 Ti4600 Benchmarks at GameSpot

Rand

Lifer
Oct 11, 1999
11,071
1
81
GameSpot has a comparison between a bunch of OEM systems... the usual crap, nothing particularly interesting... EXCEPT they also have benchmarks on one AlienWare system configured with a GF4 Ti4600, and the identical system with a Radeon 9700.

It seems to be a reg. R9700, not 9700Pro.

Benchmarks start here
 

WyteWatt

Banned
Jun 8, 2001
6,255
0
0
Wow i am not impressed at all. Even at 4x FSAA and 1024x768 unreal2003 is not playable on the ATI 9700. Not impressed at all one bit. Yes the ATI 9700 is pretty much faster with FSAA on but still its not playable in unreal2003 at 1024x768 with FSAA 4x on. I know it gets faster in RTCW and Quake 3 area with FSAA on 4x but 80 fps for the GF4 ti 4600 is still playable. I don't care about 140 fps. All i want is a minimal of 60 fps with the frame rates never going lower. Not able to do that in unreal 2003 with 4x FSAA on the ATI 9700 :(

Heck the GF4 ti 4600 even gets the extact same fps as the ATI 9700 with no FSAA in unreal 2003. Thats sad. 52.4 fps is not playable really. Even at 1600x1200 the GF4 ti 4600 is not far away from the ATI 9700. Only about 8 to 9 fps thats hardly any difference at all. But neither 29 or 38 fps is playable so why care. :(

Like i expected the ATI 9700 is overhyped and not as fast as expected. Too bad for the people who are buying it for $400 i feel bad for them. But oh well its there money.
 

Rand

Lifer
Oct 11, 1999
11,071
1
81
Well the R9700Pro should be a bit faster, and hopefully final drivers will offer a boost.... maybe that will get it up to levels at which you consider it playable.

Personally I'm less interested in the specific frame rate as I am in the relative performance boost over the Ti4600. As I have no interest in FPS (First Person Shooter) games, the specific frame rate attained means little to me.

At 1024x768 the R9700 is often only a few FPS ahead of the Ti4600, once bandwidth and graphics card limitations push to the fore-front it quickly broadens that lead as evidenced by 1600x1200 performance.

The R9700 evidently loves FSAA, often pushing 1.5-2X the Ti4600's #'s.
That doesnt necessarily mean much right now though as we have no idea as to the visual quality it can offer.
Nor have we the faintest hint as to how FSAA is implemented besides some form of MultiSampling which really only gives us a superficial impression.

Unfortunate that no DX9 apps are available to show the R9700's performance in apps that fully utilize it's capabilities.


Too bad there were only a few apps tested, and the fact that nothing besides FPS were tested is less then pleasing also.
I so wish more reviewers would notice that not everyone plays or even gives a damn about FPS.... a little more variety in testing would be a huge addition.

Ah well, at least it gives us a bit better impression of the performance of the R9700. We likely won't know much more until we see full reviews.

 

WyteWatt

Banned
Jun 8, 2001
6,255
0
0
Well i care about just having 60 fps minimum a lot of the time in most games. Except some games that you can have lower fps like racing games, plane games, etc. I am not asking for a crazy 100 to 300 fps. No way. But it seems like the ATI 9700 can't even do 60 fps at 1024x768 on unreal 2003 with all details on high or with FSAA enabled :( I really don't think its worth $400 the ATI 9700 or the ATi 9700 pro because you can't even play unreal 2003 on 1024x768 with all details on high with 4x FSAA on at 50 fps or more. Thats just not good.



 

BFG10K

Lifer
Aug 14, 2000
22,709
3,002
126
I for one am not concerned the least about what those results show and I'm waiting for the likes of Tom and Anand to put up real benchmarks.

For one thing, who knows how those OEM systems are setup and configured?
Secondly, those graphs really don't show a wide range of data.
 

AdvancedRobotics

Senior member
Jul 30, 2002
324
0
0
"For one thing, who knows how those OEM systems are setup and configured?"

Thats a good point, there is no mention as to the other parts in these systems. They could have 200MHz processors with 64Mb of SDR RAM for all we know.

My guess is it will perform much better when actual drivers are released--this test wasn't that in-depth at all.
 

hans007

Lifer
Feb 1, 2000
20,212
18
81
Originally posted by: Asuka
"For one thing, who knows how those OEM systems are setup and configured?"

Thats a good point, there is no mention as to the other parts in these systems. They could have 200MHz processors with 64Mb of SDR RAM for all we know.

My guess is it will perform much better when actual drivers are released--this test wasn't that in-depth at all.



actually, gamespot took the alienware machine, and swapped the card with a ti4600. so the systems were identical.


it was a 2.53 p4, 512mb pc1066, on an intel 850e board an audigy, and a 100 gig 7200rpm , are the relevant specs to graphics benchmarks





and alienware apparently tweaked out the system so it was a very fast gaming machine. this basically would have been the ideal platform to test on.


the actual review is

http://gamespot.com/gshw/stories/flat/0,12880,2877609-6,00.html


it shows the specs







also ati renamed the 9700 the 9700 pro, so basically this is their fastest card that will be coming out next month.
if you click on their site, under products, they only have 9700 pro, because they renamed the card from plain 9700.

 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
Originally posted by: imtim83
Wow i am not impressed at all. Even at 4x FSAA and 1024x768 unreal2003 is not playable on the ATI 9700. Not impressed at all one bit.

Heck the GF4 ti 4600 even gets the extact same fps as the ATI 9700 with no FSAA in unreal 2003. Thats sad. 52.4 fps is not playable really. Even at 1600x1200 the GF4 ti 4600 is not far away from the ATI 9700. Only about 8 to 9 fps thats hardly any difference at all. But neither 29 or 38 fps is playable so why care. :(
.

38 fps is at least twice as playable as 29 fps, I used Unreal Tournament with about 35 fps average when that first came out and found it immensely enjoyable, but anything under 30 just wouldn't do.
I think that 38 is easily useable, 60 is unnecessary need to be able to enjoy a game, 35-40 is very nice imho (maybe it's just me being brought up on not top of the range hardware)
 

Rand

Lifer
Oct 11, 1999
11,071
1
81
also ati renamed the 9700 the 9700 pro, so basically this is their fastest card that will be coming out next month.
if you click on their site, under products, they only have 9700 pro, because they renamed the card from plain 9700.


Anand and a number of other hardware reviewers have already said there will be a Radeon 9500/9700 and 9700Pro, I suppose this could have changed within the last few days but that seems dubious.
I believe it was his NV30 article that specified a core clock of 320MHz, Mem clock of 310MHz for the 9700Pro and slightly lower clockspeeds for the reg. version.
 

AnAndAustin

Platinum Member
Apr 15, 2002
2,112
0
0
:) Well the Rad9700 still seems decently faster than the 4600 for those willing to pay top 'dollar'. I agree that we need good independent reviews from AnAnd, TomsHW, TechReport, Firing Squad, X-Bit, HardOCP etc.

;) One thing which does strike me is how everybody goes on about the Rad9700's superior AA. That isn't entirely true even before factoring in things like quality, the Rad9700 has better 4xAA, comparing the 2xAA setting or QxAA on the GF4 would be a fairer comparison. Don't get me wrong true 4xAA is where most people would like to be, but as always there are plenty of options and compromises open. 800x600 with 4xAA, 1024x768 with 2xAA/QxAA or 1600x1200 without AA for example.
 

CubicZirconia

Diamond Member
Nov 24, 2001
5,193
0
71
I'm still glad I bought a geforce 4. I thought I'd really be kicking myself after the 9700 came out.
 

Zen0ps

Member
Feb 13, 2002
27
0
0
The benchmarks show the Radeon 9700 only being slightly faster than a GF4 4600 (by slightly I mean maybe 40 percent, if you consider that slightly faster)

The numbers for UT2003 seem a little low, anands own tests on this website are higher.

The processor and the video cards are not overclocked, that is something to factor in as well.

The gamespot benchies do not appear to use Anisotropic which is by far the Radeon 9700 (and 8500) strong point. It makes everything look better vs without anisotropic and there is almost no performance hit, on the GF4 you lose about half your frames if you want the visual quality of a high anisotropic (Which is why if you want both Anis & AA the Radeon can be up to 250 percent faster than a 4600)
 

Rand

Lifer
Oct 11, 1999
11,071
1
81
One thing which does strike me is how everybody goes on about the Rad9700's superior AA. That isn't entirely true even before factoring in things like quality, the Rad9700 has better 4xAA, comparing the 2xAA setting or QxAA on the GF4 would be a fairer comparison.

I don't think anyone can claim the Radeon9700 has superior or worse FSAA. All we know is that it's reasonably fast with FSAA, and that it uses some form of MultiSampling.
How can you say with a certainty it has better 4X FSAA when we have yet to even see it's FSAA, and we have only the vaguest hints on it's implementation?
For all we know it could be akin to nVidia's Quincunx... a scenario in which some absolutely love it for it's high performance and efficient removal of jaggies, and others absolutely despise it for killing texture quality.

At this point it's ridiculous to say anything other then the fact that it seems to be reasonably fast. Off-hand I can think of a good half dozen proposed implementations of MSAA in hardware, all of which can vary significantly in many important respects.
 

ai42

Diamond Member
Jun 5, 2001
3,653
0
0
I've been and QuakeCon the past couple of days with a few dozen working 9700s around the place. And I am utterly impressed. From my understanding from unofficial benchies from touney players and what not the 9700 quite possibley has the potential to easily beat any 4600 by a factor of two.... let me explain.

At least in the same maps/benchies whatever you get almost the same fps no matter how much eye candy you turn on. So around 90s fps in 3d Mark 2001 nature scene (default settings not FSAA etc), and still in the 90s with FSAA and everything turned on to max settings. And at the HardOCP seminar Kyle ran the full benchmark (well at least the ones to give you a 3dmark score) one with nothing on and got 14k and then a second time with FSAA and everything turned on and actully pulled another 100pts higher on the 3d mark score!

So that tells us there are CPU limitations since the fps are almost identical, the CPU cannot push enough power behind the 9700 to squeeze off more fps. Given these are top of the line CPUs too! Tourny computers are 2.53 Ghz P4s and Kyle's computer was an undisclosed AthlonXP (presumably the 2600 or 2800).

However we know that this not the case for the 4600.

I don't think we will know the full power of the 9700 until it hits retail and somebody from Japan cryogenically cools a P4 to 7Ghz+ with a 9700 and runs some benchies.