ATI Responds to NVIDIA's GeForce 7950GX2

[onethreehill]

"ATI however, has come forward to say that new GeForce 7950GX2 boards provides almost no improvement over the previous generation, all while costing a hand and leg. In fact, ATI goes to show that in many performance benchmarks -- synthetic and real-world -- NVIDIA's GeForce 7950GX2 actually loses to a single Radeon X1900XTX in the majority of tests."
http://www.dailytech.com/article.aspx?newsid=2735

[cooter]

Thank god someone finally told nvidia enough already......

if someone doesnt stop them that 8 card post i seen is actually going to come true.....

[Jester FPS]

great post!

[Ominous Gamer]

The ATI PR department is a joke.

First off the graphs are morphed, there is no baseline.
Secondly its percentages. So while the FPS difference is a killer 5 frames, holy crap, look at the grap they can create... Each bar set uses a different measure as well because each group sets to 100%, yet each number is different.

Not to mention that in the first graph, the cards aren't even set to like standards. Or the fact that it took ATI forever to get their drivers for X1900XTX to score like they are, yet nVidia just released this product.

Also, its odd how ATI wins in ATI tests, but as the article shows, no one else gets these results.

[Shadowmage]

^^ Stop being biased and actually look at the graph. There are numbers above each bar.

X1900XTX performance has been great from the start. Keep in mind that sites like Anandtech compared the X1900XT to the 7950GX2, and even the XT did decently at resolutions up to 16x12

This graphing style is very popular for comparison purposes, and for good reason. As long as ATI gives numbers, such graphing styles are actually preferred.

I'm just not sure about the 2nd graph... it's 2560 and 4xAA, with the 7950GX2 at default quality. 7950GX2 should be doing better in that case.

[Ominous Gamer]

^^ Stop being biased and actually look at the graph. There are numbers above each bar.

X1900XTX performance has been great from the start. Keep in mind that sites like Anandtech compared the X1900XT to the 7950GX2, and even the XT did decently at resolutions up to 16x12

This graphing style is very popular for comparison purposes, and for good reason. As long as ATI gives numbers, such graphing styles are actually preferred.

And a product of ATI PR isn't biased? What are you smoking?
ATI apparently doesn't even know how to load up 3dmark06, didn't seem to be a problem for other sites...




Also, ATI is the only place that releases benchmarks like this. Even [H] with their "best playable settings" turns its nose up to an unbalanced display like this. Its one of the first things you are taught to watch for and avoid in statistics classes, because its so misleading.

EDIT: ATI would have been more believable if they put out honest results, in a proper format; and made it a price/performance argument.

[Shadowmage]

That's X1900XTX 6xAA/16xAF vs 7950GX2 8xAA/16xAF HQ driver settings (to prevent shimmering).

Having both a X1900GT on the desktop and a 7800GTX Go on my laptop, I can attest to these settings being IQ-wise close to identical.

As I also said, the first graph looks about right. 8xAA is just about unplayable on any NVIDIA card, since it's supersampling. NVIDIA cards are also usually tested on Q, not HQ.

I'm not sure about the 2nd graph, as I also said b4.

Your 3dmark06 graph is pointless since it's 1280x1024, not the settings above. Same with the 2nd graph. Settings aren't the same.

Obviously ATI wants what's best for ATI (hence biased) but I see nothing incorrect with the first graph.

[einCe]

dont forget the price gap between these

[OmegaMerc]

Don't forget there is ALWAYS damage control after you release a new card for the opposite team.

[b0bd0le]

And a product of ATI PR isn't biased? What are you smoking?
ATI apparently doesn't even know how to load up 3dmark06, didn't seem to be a problem for other sites...




Also, ATI is the only place that releases benchmarks like this. Even [H] with their "best playable settings" turns its nose up to an unbalanced display like this. Its one of the first things you are taught to watch for and avoid in statistics classes, because its so misleading.

EDIT: ATI would have been more believable if they put out honest results, in a proper format; and made it a price/performance argument.

apparently your stats class didn't teach you how to read graphs

ati ran 3dmark & call of duty at different settings

are you sure you're not blind?

[ahmad]

apparently your stats class didn't teach you how to read graphs

ati ran 3dmark & call of duty at different settings

are you sure you're not blind?

Not only that but he is looking at COD2 numbers at 4xAA. 4xAA with a $700 piece of hardware? A 7600GT could play games at that

[Bar81]

comparing ATI and nvidia at the "same" settings is a farce. nvidia doesn't produce vid quality even close to ati at the "same" settings.

The GX2 is a joke. I only hope the usual idiots are not dumb enough to buy this and encourage nvidia.

[BlackX]

I only hope the usual idiots are not dumb enough to buy this and encourage nvidia.

Unfortunately they are..

[NickS]

Yeah, the GX2 is a joke... a complete joke. But, it's very, very fast in SLi . It's pretty cool though that an X1900XTX with better image quality can keep right up with it in most of the tests though .

Nick

[mdzcpa]

LOL...since when did anyone seriously take any performance results from the mouth of the manufacturer? ATI's PR graphs are a joke. Not only in layout, but in settings used to distort the results in selected benches. Those that defend it must be fanboys I guess. I've never understood the need to defend a brand....its odd behavior to me.

Anyways, until someone else on this threads owns, benches, and plays with BOTH cards (X1900XT and GX2) like I have, there simply won't be much credibility to the opinions offered. The GX2 CRUSHES the X1900XT in every single game and bench I've thrown at it. Sorry ATI fanboys, but I've put my X1900XT back in its box to sell.

Here's some graphs from an unbiased site (Tech Report):










Not to mention the GX2 is quieter and consumes less power than the X1900XT:


Tech Report Conclusion:
"This card takes up no more space, draws no more power, and generates no more heat or noise than a Radeon X1900 XTX, but its performance is in another class altogether."

Its hard to read every independent review and not draw the same conclusion. There isn't a review out there that doesn't have the GX2 out performing the X1900XT with less power consumption, and less noise.

Don't get me wrong, my X1900XT has serious merits. IQ is definitely great. But, the GX2 is simply much better for only $100 more. Where the X1900XT would choke playing Oblivion at high detail settings, the GX2 churns things out buttery smooth.

I only call it as I see it with my own two hands folks.

[NickS]

Jesus christ, my X1800XT consumes more power than a 7900 GTX? LOL.

But look @ those graphs, the X1900XTX is cheaper and has better 'graphical quality' and it keep's up with Dual GTX's (7950GT2) quite nicely . It is pretty amazing though, the way that the GT2 draws as much power pretty much and runs quieter.

Nick

[Bar81]

Jesus christ, my X1800XT consumes more power than a 7900 GTX? LOL.

What an asinine review. Yeah, I'm sure the X1800 XT consumes nearly as much power as the two cores on the GX2

[Victorshen]

refer to my sig pls

[mdzcpa]

Yeah, I'm sure the X1800 XT consumes nearly as much power as the two cores on the GX2

Its a fact. The GX2 is run at reduced power levels, this is why it could never keep up with a pair of 7900GTXs in SLI, or X1900XTs in CF.


EDIT- BTW, although I will allow some spirited debate regarding ATIs "PR Response" I will not let this thread degenerate into a fanboy flamefest. That goes for both sides. Stick to the facts and stow the flames. Thanks.

[Bar81]

It's running at 500mhz x2. There's no freaking way no matter what process enhancements it has versus the X1800 XT.

edit:

here you go in a comparison with the much more power hungry X1900 XT
http://www.anandtech.com/video/showdoc.aspx?i=2769&p=3

[OmegaMerc]

I ordered a 7900gx2 and reading this thread is gonna make me jump off a ing building I swear .

[Starscream]

What an asinine review. Yeah, I'm sure the X1800 XT consumes nearly as much power as the two cores on the GX2

wake up and use google.
every review site that tests power usage comes to the same conclusion.
Sadly most sites dont test power consumptiopn but with a little googling you can get alot of info.
Contrary to what alot of people believe the X1800 and X1900 use alot of power.

To many people swallow ATIs PR stuff as the truth and cause ATI nvr includes power usage info in them people (usually fanboys) dont care about it.

bottom of the link:
http://tweakers.net/reviews/628/2

[Bar81]

So you're saying that the X1800 XT and X1900 XTX consume close to the same power:

http://www.anandtech.com/video/showdoc.aspx?i=2679&p=15

There's a freaking 60 WATT power difference. Tech report (as usual) is amateur hour.

Thus ends another episode of buy a clue...

[Xenogias]

why do people care about power consumption anyway? ^_^

[mdzcpa]

It's running at 500mhz x2. There's no freaking way no matter what process enhancements it has versus the X1800 XT.

edit:

here you go in a comparison with the much more power hungry X1900 XT
http://www.anandtech.com/video/showdoc.aspx?i=2769&p=3

Thanks for the link! Those results you linked are right in line with what was posted by the Tech Report review. I see at Anand's that the GX2 still draws less power than the 1900XT under load. The only difference between Anand's and Tech Report is that the TR used the 1900XTX which accounts for even higher power draw by ATI. Anand's text commentary addresses that as well.

So you're saying that the X1800 XT and X1900 XTX consume close to the same power:

http://www.anandtech.com/video/showdoc.aspx?i=2679&p=15

There's a freaking 60 WATT power difference.

Power usage information is ALWAYS measured relatively. Tester to tester will always show different power measurements. What you are supposed to draw from the graphs are power usage relative to the other cards, not hard and fast wattage measurements. Individual system specs greatly effect the measurements.

Here's an example. Look at the two Anand's reviews linked. Each review has different power usage under load for the X1900XT. (276w vs 319w)





Does this make Anand's the "amatuer hour" ?

If you look at the Tech Report, they are showing the difference between the X1800 and the X1900XTX to be 34watts. BUT, the total wattages measured are down compared to Anand. However if you look at the differences between power used by card in percentage. Its about the same.

TR:
X1800XT 230
X1900XTX 264
Difference 34
As % 15%

Anand's:
X1800XT 282
X1900XTX 341
Difference 59
As % 21%

That's easily close enough to draw accurate relative power usage information.