ATI Crossfire Beats Nvidia SLI

Turning off optimisations for the 7xxx range closed the IQ gap considerably, while I would agree even then that ATI had the "crown" when it came to IQ, with optimisations turned off I would also say that the difference was negligable. Texture crawling or "Shimmering" was a bit of a problem for some people but using the clamp optinions seemed to fix that for the most part.

turning off the optomizations also took a considerable hit in FPS making it lag behind the 19xx series a noticible amount :p and it still had SOME crawling even without optimizations on
 
turning off the optomizations also took a considerable hit in FPS making it lag behind the 19xx series a noticible amount :p and it still had SOME crawling even without optimizations on

That's not true. The performance hit was no more than 3-4 fps. Either way, the X18xx/X19xx did have the IQ advantage over GeForce 78xx/79xx cards. That changed with G80, since it set a new "standard" for image quality.
 

You do realize this card was released in what.. April of this year?

Another great way to show off the size is to compare it to a "standard" sized graphics card, like the GeForce 7900 card seen below. The Radeon X1950 Pro Dual simply dwarfs this other card. The Radeon X1950 Pro Dual is, in comparison, also quite a larger than GeForce 8800 and GeForce 7950 GX2 graphics cards.

Its LARGER than an 8800 GTX card, and trades blows with the 320mb 8800 gts. Yet it can't do DX10 and costs more. (Size Comparison: http://www.hexus.net/content/item.php?item=8282&page=2). It also gets spanked in OpenGL games (Q4) and when HDR+AA is used.

Here it is at newegg: http://www.newegg.com/Product/Product.aspx?Item=N82E16814102096 only $419 + shipping too :rolleyes:
 
This card should be compared to the GX2 because it is two cards on a single slot just like the GX2. Compairing the GX2 with other single cards is lame because the GX2 is not a SINGLE card :rolleyes:
 
The 7950GX2 is a single video card. When you buy one, you get one card in the box. The two PCBs are in-seperable. It doesn't matter how it's built or what it is. The fact is it's a single self-contained unit. It is totally proper to compare it to a single X1950XTX. Just as you would rightfully compare a X1950XTX Crossfire system against a 7950GX2 SLI system. Each card represented the pinnacle of what NVIDIA and ATI had to offer at the time. It's one fucking card, stop splitting hairs.

All that matters is that when you bought a single 7950GX2 you were getting a single card that beat the shit out of anything ATI had to offer 90% of the time and Quad-SLI beat the shit out of any dual ATI video card configuration. These are the facts, this is what matters. How the 7950GX2 was constructed and what made it work are not relavant.

The fact remains that NVIDIA has been on top of ATI for some time now and this fact doesn't look like it will be changing anytime soon.
 
This card should be compared to the GX2 because it is two cards on a single slot just like the GX2. Compairing the GX2 with other single cards is lame because the GX2 is not a SINGLE card :rolleyes:

Wrong, all cards should be compared to products in their price / performance / feature range. You are trying to say that we can only compare 512mb cards to each other because they both have 512 ram.

Who cares how much ram a card has (or if its "two in one"). We care about price, performance, and what features it offers.
 
Wrong, all cards should be compared to products in their price / performance / feature range. You are trying to say that we can only compare 512mb cards to each other because they both have 512 ram.

Who cares how much ram a card has (or if its "two in one"). We care about price, performance, and what features it offers.

Exactly right.
 
The 7950GX2 is a single video card. When you buy one, you get one card in the box. The two PCBs are in-seperable. It doesn't matter how it's built or what it is. The fact is it's a single self-contained unit. It is totally proper to compare it to a single X1950XTX. Just as you would rightfully compare a X1950XTX Crossfire system against a 7950GX2 SLI system. Each card represented the pinnacle of what NVIDIA and ATI had to offer at the time. It's one fucking card, stop splitting hairs.

All that matters is that when you bought a single 7950GX2 you were getting a single card that beat the shit out of anything ATI had to offer 90% of the time and Quad-SLI beat the shit out of any dual ATI video card configuration. These are the facts, this is what matters. How the 7950GX2 was constructed and what made it work are not relavant.

The fact remains that NVIDIA has been on top of ATI for some time now and this fact doesn't look like it will be changing anytime soon.

I agree with you there but some points I would like to make. The 7950GX2 SLI setup was billed as a Quad SLI setup, from big name computer vendors as well even Nvidia had it advertised as such on there website. Its pretty confusing since its really a single card with 1 PCi-E connection even though its 2 PCB's and memory/GPU on each, its pretty easy to confuse something like this considering the vendors these days use advertising to make everything massively confusing.
 
I always thought that the x1900xtx faired better in current games then the 7900gtx. But you could skew benchmark results to prove either card was "better"

The fact remains that NVIDIA has been on top of ATI for some time now and this fact doesn't look like it will be changing anytime soon.

Frankly I don't think having an inferior GPU is what has/is holding ATi back. Their problem is they can't deliver a platform.

For a second let's drop the debate of if crossfire is better then SLI or not.

When how many real crossfire boards were there for socket 939, not many, and they weren't easy to find. Then think of how many SLI boards there were, nforce4 was all over the place. The same thing happened on socket AM2.

The only saving grace for crossfire was intels supporting it, but even then it's not true x16/x16. And what about ATis own board, only DFI offers an actual solution here.

What's the point in having a dual GPU solution you can't get to market?

Releasing a product late to market and in limited quantity seems to be par for the course with them.
 
Wrong, all cards should be compared to products in their price / performance / feature range. You are trying to say that we can only compare 512mb cards to each other because they both have 512 ram.

Who cares how much ram a card has (or if its "two in one"). We care about price, performance, and what features it offers.

Yet people still say the HD2000 series is a failure eventhough the cards are priced nicely according their performance with added features like HDMI audio and UVD :rolleyes:

The 7950GX2 is a single video card. When you buy one, you get one card in the box. The two PCBs are in-seperable. It doesn't matter how it's built or what it is. The fact is it's a single self-contained unit. It is totally proper to compare it to a single X1950XTX. Just as you would rightfully compare a X1950XTX Crossfire system against a 7950GX2 SLI system. Each card represented the pinnacle of what NVIDIA and ATI had to offer at the time. It's one fucking card, stop splitting hairs.

All that matters is that when you bought a single 7950GX2 you were getting a single card that beat the shit out of anything ATI had to offer 90% of the time and Quad-SLI beat the shit out of any dual ATI video card configuration. These are the facts, this is what matters. How the 7950GX2 was constructed and what made it work are not relavant.

The fact remains that NVIDIA has been on top of ATI for some time now and this fact doesn't look like it will be changing anytime soon.

Can you use more than one monitor with 7950GX2 running as a single card?
 
All that matters is that when you bought a single 7950GX2 you were getting a single card that beat the shit out of anything ATI had to offer 90% of the time and Quad-SLI beat the shit out of any dual ATI video card configuration. These are the facts, this is what matters. How the 7950GX2 was constructed and what made it work are not relavant.

The fact remains that NVIDIA has been on top of ATI for some time now and this fact doesn't look like it will be changing anytime soon.


http://enthusiast.hardocp.com/article.html?art=MTE0NCw1LCxoZW50aHVzaWFzdA==
http://enthusiast.hardocp.com/article.html?art=MTE0NCw2LCw3MA==
http://enthusiast.hardocp.com/article.html?art=MTE0NCw3LCxoZW50aHVzaWFzdA==

etc etc

infact the best dual card ATI solution won more games then the Quad SLI 7950GX2

damn don't you read [H] Reviews :p?!
 
Yet people still say the HD2000 series is a failure eventhough the cards are priced nicely according their performance with added features like HDMI audio and UVD :rolleyes:

Priced nicely? Hmm.... I fail to see how a $400 card giving less performance than a $250 card is "Priced Nicely".

http://www.hardocp.com/article.html?art=MTM1MSwsLGhlbnRodXNpYXN0

As for the rest of the cards, they are crap for gaming, their only saving grace is the fact that they are for home theater systems.

The low end 8 series also provide a good home theatre solution and can also be used for some gaming.
 

Those reviews seem to show both cards running @ about the same settings and FPS, with the nVidia card having a lead in fps for many. Not having an in between for 4xAA and 8xAA is what put it behind in quality vs the 1900/1950. Now I don't know what the price was for both back then, so if the ATI card was cheaper then yes, it was a better bang/buck as performance was so close between the two.
 
Rest of the cards are crap because they are dirt cheap. Like you said look at the price/ performance/ feature range. I agree that HD2900 is overpriced but I don't think that 320MB is enough for high resolution.
 
The 2000 HD series cards will really shine with a revision. I can see them with a revision beating Nvidia and being the best choice. They are decent cards now , just power hogs / vaccum cleaner fans are what holds them back the most.

Most folks wont upgrade there PSU in order to upgrade there video cards. They want something they can just add and start playing games asap. Right now ATI's card just isn't effective enough considering its requirements. A buddy of mine recently bought a 2900XT and he has a 450 watt PSU , he's running a Core 2 duo , 2 hard drives and audigy 2 . He installed it correctly (I checked) and has the PCi-E power adapters connected to seperate lines directly to the PSU. The moment he started it up it just starts up and shuts off with a loud beep. He asked me if there was something wrong with his card, so I brought his card over to my place and installed in my PC (I have a 700 watt thermaltake) and hooked it right up , booted just fine no issues.

So he went out and got a 620 watt Corsair PSU (seasonic PSU in pretty dressing) and installed it. Now everything works fine for him but in the end his single card upgrade cost about 589 dollars. Not really alot cheaper than a 8800 GTX.
 
Rest of the cards are crap because they are dirt cheap. Like you said look at the price/ performance/ feature range. I agree that HD2900 is overpriced but I don't think that 320MB is enough for high resolution.

So get a (still cheaper) 640mb card... which.. omg, has more memory than the 2900 :rolleyes:
 
So get a (still cheaper) 640mb card... which.. omg, has more memory than the 2900 :rolleyes:

This is what this thread is all about, how would a CF HD2900 system compare with a SLI 8800GTS 640MB system or a single 8800Ultra system.
 
This is what this thread is all about, how would a CF HD2900 system compare with a SLI 8800GTS 640MB system or a single 8800Ultra system.

Actually, this thread was about a test which used a 2900 xfire system vs a 8800gtx sli system in which they removed all of the results where the 8800gtx system won, and used buggy drivers which only basically used 1 gtx in vista, not SLI.
 
Those reviews seem to show both cards running @ about the same settings and FPS, with the nVidia card having a lead in fps for many. Not having an in between for 4xAA and 8xAA is what put it behind in quality vs the 1900/1950. Now I don't know what the price was for both back then, so if the ATI card was cheaper then yes, it was a better bang/buck as performance was so close between the two.

did you even read the review? look at the XF the CF 1950XTX was ahead by a fair bit, 10-15fps with higher settings in some cases and it was cheaper =p
 

I did read those reviews. In fact I owned two X1950XTX's in Crossfire at the time. Mainly though I purchased them because they were cheaper than the 7950GX2's and Quad-SLI wasn't all that great in terms of game support. In any case the general consensus is that the X1950XTX was slower than the 7950GX2. I believe this to be generally true.
 
The 2000 HD series cards will really shine with a revision. I can see them with a revision beating Nvidia and being the best choice.

Well they said the rest of the HD 2000 series would shine due to being 65nm parts. Look at how accurate that line of thinking ended up being.
 
ALL Nvidia vs ATi threads degenerate fast.

WTF are ya'll arguing over last gen?

So tired of this I almost wish Ati would just fold so there would be no more debate.
 
did you even read the review? look at the XF the CF 1950XTX was ahead by a fair bit, 10-15fps with higher settings in some cases and it was cheaper =p

http://enthusiast.hardocp.com/image.html?image=MTE1NTc4MTg0MkN1Q1k5b0Z1MVpfNV81X2wuZ2lm

ATI: Better AF (8xHQ vs 8xNv)
Nvidia: Shadows Enabled, Plus higher FPS, almost 10fps higher on avg.

Prey was about the same for both, NV was ahead in "single card" configs and tied at XF/SLI.

I mentioned the rest in my earlier response.
 
http://enthusiast.hardocp.com/image.html?image=MTE1NTc4MTg0MkN1Q1k5b0Z1MVpfNV81X2wuZ2lm

ATI: Better AF (8xHQ vs 8xNv)
Nvidia: Shadows Enabled, Plus higher FPS, almost 10fps higher on avg.

Prey was about the same for both, NV was ahead in "single card" configs and tied at XF/SLI.

I mentioned the rest in my earlier response.

you are comparing 1 out of the 4 games that were tested, look at the other 3, plus, I was talking about Dual Card solution, as in SLI 7950GX2 Vs CF 1950XTX :p I know single card the XTX loses, but dual card it wins most of the time :) AND is cheaper
 
turning off the optomizations also took a considerable hit in FPS making it lag behind the 19xx series a noticible amount :p and it still had SOME crawling even without optimizations on

Nah, performance hit was quite small actually, not enough to suddenly give ATI the lead in performance, setting the drivers to high quality mode (which also happened to force off all the optimisations) caused a large hit in performance which is probably what you're thinking about.

If you were overly concerned about IQ then no doubt ATI was the way to go but for the most part I dont think the differences in IQ with all the optimisations off was enough for the average gamer to notice.

Anyhow I don't want to digress, but I feel the point should be made.

Back on topic this review is clearly another in the line of several dodgy reviews which seems to swing all of their results in favour of ATI to make the 2900 appear better than it is, clearly if 1 2900 cannot beat a 8800GTS then 2 in crossfire isn't going to beat 2x 8800GTX in SLI.

I've never seen any well documented evidence to show that either Crossfire or SLI is inherently more efficient at combining graphics cards, certainly not by a large margin, which instantly sends up a red flag about any review like this.
 
I won't say that the GX2 crushed everything ATi had, it is faster than everything ATi had but it should be faster because it is two cards in a single slot so it is not a fair comparison if you campare it with an ATi's single card. The GX2 should only be compared with Sapphire Dual X1950Pro. If you want to compare the GX2, you should at least compare it with a Crossfire setup because the GX2 is two 7900GT in one slot.

Newsflash: There is no "fair" in the industry other than that determined by the performance offered by a particular price range, regardless of whether its 1, 2, or 16 gpu's. If ATI could've matched it they would have. They didn't sit there and think "oh lets hold off to be fair". It's the business world.
 
HTF did 8800GTX in SLI manage to only get 12k in 3DMark06!?!?!

I can get 11805 with 1 GTX and a dual-core CPU!!!

lolz
 
HTF did 8800GTX in SLI manage to only get 12k in 3DMark06!?!?!

I can get 11805 with 1 GTX and a dual-core CPU!!!

lolz

for one, either SLI sucks in 3dmark, or its vista with broken sli ?

SLI in 3dmark 2k6 only yields 2-3k increase usually =)

people with sli getting 13-14k on normal systems and 11ish one single card
 
Yeah, I just said that since I've seen quite a few 17k-18k from SLI.
 
Yet people still say the HD2000 series is a failure eventhough the cards are priced nicely according their performance with added features like HDMI audio and UVD :rolleyes:



Can you use more than one monitor with 7950GX2 running as a single card?

Yes.
 
HTF did 8800GTX in SLI manage to only get 12k in 3DMark06!?!?!

I can get 11805 with 1 GTX and a dual-core CPU!!!

lolz

Because it's a stupid synthetic benchmark that doesn't perform very well on 8800GTX SLI systems. Also, CPU speed makes all the difference in 3D Mark. You can jump up thousands of points simply by overclocking your CPU. Does that make sense? It doesn't to me considering CPU speed has a minimal impact on games past certain clock speeds on each architecture.

In the real world, the 8800GTX/Ultra SLI systems are the fastest and most powerful computer gaming platform.
 
Yeah - I get you about 3DMark06 being synthetic, and not performing well with GTX's in SLI...
however I've just seen a lot of GTX's in SLI get 16k - 18k in in 3DMark06.

As it goes for the CPU speed it was @ 3.7 and it was a quad!!! So considering that I get ~ 10K with all settings at stock with one GTX and only 2 cores, so it just makes that score very unbelievable for me... that's all
 
Yeah - I get you about 3DMark06 being synthetic, and not performing well with GTX's in SLI...
however I've just seen a lot of GTX's in SLI get 16k - 18k in in 3DMark06.

As it goes for the CPU speed it was @ 3.7 and it was a quad!!! So considering that I get ~ 10K with all settings at stock with one GTX and only 2 cores, so it just makes that score very unbelievable for me... that's all

That's the thing I noticed. With my E6600 at stock speeds my scores were horrid. When I overclocked my processor to 3.6GHz I was able to get 16k no problem.
 
Don't know if this is new news or old news, but I had never seen this before:

http://www.falcon-nw.com/keltstake/ati_2900xtx.asp
Since Falcon NW overclocks the CPU so much, do they also overclock the video cards, and/or tweak the drivers?

Something is going on there that didn't show on the other reviews, strange because a print magazine has a long lead time and AMD's dancing in the aisles drivers would have come out too late.
 
Since Falcon NW overclocks the CPU so much, do they also overclock the video cards, and/or tweak the drivers?

Something is going on there that didn't show on the other reviews, strange because a print magazine has a long lead time and AMD's dancing in the aisles drivers would have come out too late.

If I understood it right, the main reason would be that these are not "stock" 2900XTs nor even the pre-release 1GBDDR4 2900XTs, but rather a cherry-picked model only available to integrators--that plus the driver issues in Vista. So their results would not compare to those in other reviews, as the ATi cards are in some respects a different animal.
 
Since Falcon NW overclocks the CPU so much, do they also overclock the video cards, and/or tweak the drivers?

Something is going on there that didn't show on the other reviews, strange because a print magazine has a long lead time and AMD's dancing in the aisles drivers would have come out too late.

I notice something real different.
Manny claim that bench is flawed. I think not.
Because it's a Vista 32 bench. ifso nV is so great, its great in XP but suck in Vista with SLI. So the reviewer recommend nV SLI for XP and CF for Vista.

So where ATI have a lot to do driver wise for XP and Vista.
nV had a lot to do for Vista.

So whatever 3dmark score you get in win XP those scores aren't relevant with this bench test.

What this review shows if you go with HD2900XT in CF it could be the fastest in DX10 with crysis on Vista. Until nV drivers solve the SLI scaling in Vista. In time.
 
Back
Top