Xbit releases massive graphics card roundup

[GeneralGrievous]

Enjoy

Quite possibly the best presented, most comprehensive test I have seen. Many thanks to xbit for doing what must have been a whole lot of work. The graphs at the end are especially useful. Something to note is the abysmal performance of the 6800 nu, and the game "perimeter" which gets 35 fps on an x800 XT at 1024x768 pure speed.

 

[Schadenfroh]

why does the 6800ultra and the x800XT PE get 55/68 FPS in deus ex 2 At 1024x768 with no af or fsaa? seems like they would be owning that game. does the coding for it suck or something? even with the pure speed settings

 

[Actaeon]

Wow!

Thanks for the link! I was looking for the performance differences between the 6800GT and the Ultra, to see if it was worth overclocking or not.

Hopefully this answers my question . Plan on ordering a 6800GT tomorrow.

 

[gururu]

thanks Generalgrevous! a good read.

 

[Marsumane]

Another set of benchmarks to argue/agree with the millions that ive seen already...

 

[OnEMoReTrY]

Wow, the X800 Pro wasn't killed nearly as bad as people make it out to be against the GT. The fact that the average pro OC's so far compared to the GT is a great sign. I got mine to 580/580 on stock cooling with no volt mod's. Run's perfectly stable. Not a bad showing at all!

 

[BFG10K]

does the coding for it suck or something?

DE2 is using per-pixel lighting and shadowing like Doom III but the implementation is quite slow.

 

[Lonyo]

Originally posted by: Schadenfroh
why does the 6800ultra and the x800XT PE get 55/68 FPS in deus ex 2 At 1024x768 with no af or fsaa? seems like they would be owning that game. does the coding for it suck or something? even with the pure speed settings

The coding sucks for it.
It's a dog, and it has always been a dog.

 

[Lonyo]

Why do they have 8xAF/16xAF listed on quality benchmarks?
Does one lot of cards have 16x and one 8x? Or is it max for each card (which can be 8x or 16x)?


And drivers were Cat 4.6's, whcih offer reduced speed, IIRC, compared to the 4.5 and 4.7's, but only by around 5% in most cases.

 

[Nebor]

Originally posted by: Schadenfroh
why does the 6800ultra and the x800XT PE get 55/68 FPS in deus ex 2 At 1024x768 with no af or fsaa? seems like they would be owning that game. does the coding for it suck or something? even with the pure speed settings

I bought it the day it came out, I was so psyched, continuing the Deus Ex saga. I took it home to my then top of the line p4 3ghz @ 3.5ghz, and r9800 pro @ 420\360.... And was presented with a wholly unplayable game. It was literally a slideshow. The forums for the game were swarmed with pissed off people. The game was a poor xbox port to the PC. I'm amazed that even the new powerhouse cards can run it decently, if only at 10x7

 

[Gamingphreek]

Nice review however it is very slanted towards old ATI cards. I find it hard to believe that a 9600XT outperforms all the 59xx series cards in a lot of tests. Also i find it hard to believe that the 5700U is performing on par with the 5900's.

Something seems wrong.... however the next gen cards seem to be benched very accurately.

Thanks for the read general.

-Kevin

 

[kylebisme]

Originally posted by: Gamingphreek
Nice review however it is very slanted towards old ATI cards. I find it hard to believe that a 9600XT outperforms all the 59xx series cards in a lot of tests. Also i find it hard to believe that the 5700U is performing on par with the 5900's.

those are all 4 pipe cards. sure the 5900 has a 256 bit bus and an extra texture unit per-pipe, but if bandwidth or multitexturing isn't the botteneck then the 5900 shows it's weakness. this has been comon knowlage for some time amongst supposed "ati fanboys."

 

[Gamingphreek]

But still it should lead boith of those cards by a much larger margin, as demonstrated in the AT reviews.

-Kevin

 

[apoppin]

Originally posted by: Nebor

Originally posted by: Schadenfroh
why does the 6800ultra and the x800XT PE get 55/68 FPS in deus ex 2 At 1024x768 with no af or fsaa? seems like they would be owning that game. does the coding for it suck or something? even with the pure speed settings

I bought it the day it came out, I was so psyched, continuing the Deus Ex saga. I took it home to my then top of the line p4 3ghz @ 3.5ghz, and r9800 pro @ 420\360.... And was presented with a wholly unplayable game. It was literally a slideshow. The forums for the game were swarmed with pissed off people. The game was a poor xbox port to the PC. I'm amazed that even the new powerhouse cards can run it decently, if only at 10x7

the patch really helped.

i can get (now - barely) satisfactory FPS at 10x7 (9800xt/P4 @3.31Ghz/1GB PC3500) . . . but it DOES look "nice" and is playable . . . (however it is on a backburner as THief: Deadly Shadows is not only great looking (for a thief game) but has the BEST gameplay and STORY (in any game for a LONG time).

 

[Lonyo]

The 6800 does look like a dog in many tests, running slower than the 5900 series cards often, which is quite suprising.

 

[nRollo]

Hmmm I can't even look at that review. I get weird MYSQL fetch errors.

But I'm sure the 6800 is slower than any other card ever made, and 9600XT own 5950s, etc.. All the other review sites are dead wrong we must only trust XBit! LOL

 

[SickBeast]

Originally posted by: Gamingphreek
Nice review however it is very slanted towards old ATI cards. I find it hard to believe that a 9600XT outperforms all the 59xx series cards in a lot of tests. Also i find it hard to believe that the 5700U is performing on par with the 5900's.

Something seems wrong.... however the next gen cards seem to be benched very accurately.

Thanks for the read general.

-Kevin

I dunno, the last time an Xbit graphics article came out, people were complaining that they were slanted towards nVidia. Even in this one, if you read their reccomendations, they reccomend the 6800U over the X800XT and the 6800GT over the X800PRO. To say that they're biased is a little silly IMO. As TheSnowman said, NV30 had 4 pipes for the most part which gave them poor performance in many games.

 

[nRollo]

General:

Something to note is the abysmal performance of the 6800 nu,

WTH are you talking about? Bear in mind everyone knows the 6800NU doesn't do 16X12 due to memory limitations?

Highly Anticipated DX9 game 1
12X10 SF Demo:
6800NU:62fps X800XT:65fps

Even the 16X12 at 52 splits the difference between a 9800XT (42) and a X800P (59)?

12X10 U2 Demo:
6800NU 95fps X800Pro 103fps pretty abysmal performing at the same level as a card that costs $100 more?

How about highly anticipated game 2?

12X10 Chernobyl
6800NU 138fps X800Pro 153 fps so you must want it to be $100 less and faster?

Oh well, talk smack about the NU all you like. It's a great deal for $300. For me, if Buyxtremegear doesn't ship my XFX GT today as promised, they get the cancellation and EVGA get the order for an Ultra and I wait.

 

[kylebisme]

Originally posted by: Gamingphreek
But still it should lead boith of those cards by a much larger margin, as demonstrated in the AT reviews.

-Kevin

like i said; "but if bandwidth or multitexturing isn't the botteneck then the 5900 shows it's weakness." so it depends on what you benchmark, and AT benchmarking has had a curious tendacy avoid the weak spots in the fx architecture.

 

[GeneralGrievous]

My mistake. Looks like xbit borked the graphs at the end on page 53. They graphed x800 Pro data and the 6800 was well behind in the 1280 and 1280 AA/AF.

http://www.xbitlabs.com/misc/picture/?src=/images/video/graphics_cards_1H-1004_charts/perf_mainstream_1600_pure.gif&1=1


But your same argument can be used in reverse.

Wolfenstein: ET
Look, you get 6 fps for $100 over the 9800 Pro at 1280x1024 4x/8x.

UT2004
Outdone by the 9800 Pro at the same setting.

Interestingly enough the 9800 Pro does better in many 1600x1200 benches even with 128 mb of memory.

The 6800 U got trashed by the XT at 1600x1200 4x/8x. It won a whopping 3 benchmarks, 2 of which were 3dmark tests. Oh, and FIFA 2004.

 

[Delorian]

Awesome link, cheers:beer:

 

[Compddd]

I firmly believe Warren Spector should be executed for developing the sequel to my favorite game series for that hunk of trash called the Xbox and then porting it over to PC as a second thought with piss poor coding.

 

[ioticus]

Originally posted by: Compddd
I firmly believe Warren Spector should be executed for developing the sequel to my favorite game series for that hunk of trash called the Xbox and then porting it over to PC as a second thought with piss poor coding.

I agree, but I think he should be tortured first, then executed.

 

[otispunkmeyer]

Originally posted by: Lonyo
Why do they have 8xAF/16xAF listed on quality benchmarks?
Does one lot of cards have 16x and one 8x? Or is it max for each card (which can be 8x or 16x)?


And drivers were Cat 4.6's, whcih offer reduced speed, IIRC, compared to the 4.5 and 4.7's, but only by around 5% in most cases.

Nvidia = 8xAF and 8xAA
Ati = 16xAF abd 6xAA

 

[Keysplayr]

The only reason this thread was created was because General "liked" the results. If the results were different, this post would never have been made by him.