• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Quad CrossFire?

Sovereign

2[H]4U
Joined
Mar 21, 2005
Messages
3,098
Now that Quad SLI is officially a DIY product, the ATI crowd is probably wondering how they will keep up. One answer is the X1950XT(X) which according to the VERY ENTHUSIASTIC POST that was at the top of the forum as I type this, can beat out a 7950GX2. Congrats ATI, keep the competition up. The only way to more speed is to keep having the "other guy" come up with something faster.

Now, although I have seen triple ATI configs, this is not triple CrossFire, but rather two cards in CrossFire and one for physics....

Anyone heard anything about Quad CrossFire?
 
Personally i think this mentality of "omg slap more gpu's together for more fps!!!" is getting retarded. At the most regular sli or crossfire is more than enough for most people, then nvidia come out with their *cough* 2 card *cough* 4 gpu solution How long before we see 6 or 8 gpu's slapped together? Just seems to be getting out of hand.
 
Chris_B said:
Personally i think this mentality of "omg slap more gpu's together for more fps!!!" is getting retarded. At the most regular sli or crossfire is more than enough for most people, then nvidia come out with their *cough* 2 card *cough* 4 gpu solution How long before we see 6 or 8 gpu's slapped together? Just seems to be getting out of hand.

Agree... when you have an external AC power brick for your frickin video card there is a problem. Besides... there really isn't a need for 2 cards over 1 high end card for just about anything unless you are running stupid high resolutions.
 
Personally i think this mentality of "omg slap more gpu's together for more fps!!!" is getting retarded. At the most regular sli or crossfire is more than enough for most people, then nvidia come out with their *cough* 2 card *cough* 4 gpu solution How long before we see 6 or 8 gpu's slapped together? Just seems to be getting out of hand.

second time we've seen it happen, that was on a single card, the voodoo6000 since it was a good card, still didn't beat the competition, still this kind of mentality of reaching new performances in the industry isn't new

also how many 3d processors did the Voodoo2 card have? wasn't it like 3 physical? that in SLI would be 6, although the 3 was completely pulled out of my ass, i was too tickled happy when i got mine when they came out to really bother to check.

edit: found out it had 1 pixel, and 2 texel processors on it, so i guess the card in whole (if you think about it) was 1 3d processor
 
Well dual core CPUs are one thing, everyone seems to agree on the need for those....if GPUs could just lower their power requirements (or maybe just get them out of the case, I heard something about external PCI-E....) then we'd be fine with multicore GPUs. The next batch of procs (AM2 and Conroe, especially Conroe) are ridiculously thermally effecient compared to the previous generation.
 
I hope to god that ATI does not follow this "lead". My opinion is that nVidia cant come out with a single faster card so they just up the ante by adding GPU's. What happened to simply creating a better product. I would prefer if ATI would finally shut down this multi card thing by simply coming out with a own all card that single handedly could beat any SLI or Quad SLI rig.

In all honesty i do not hate nvidia i am just sick and tired of all this more arne more gpu stuff. when is it going to end.
 
Isn't it odd ATI and Nvidia doing multiple GPU's... 3dfx went under because of it...

Everyone was like voodoo 5 5500 was slow... It was rather slow because 3dfx basically put 2 voodoo 3's with 32 bit/AA/2048 textures and called it voodoo 5...

We're at a peak in graphics pipes and memory bandwidth...

Unless we go to 512bit memory controllers and 32/48 pipe cards we will do the multi GPU thing for a while... I'm thinking much lower die is needed to goto the next step up.. 512 bit controllers are premature....

Memory can get so much faster but it's going rather slow right now and to keep up with Nvidia and ATI 6 month rehash and whole new generation every year...
 
Well dual core CPUs are one thing, everyone seems to agree on the need for those....if GPUs could just lower their power requirements (or maybe just get them out of the case, I heard something about external PCI-E....) then we'd be fine with multicore GPUs. The next batch of procs (AM2 and Conroe, especially Conroe) are ridiculously thermally effecient compared to the previous generation.

GPU's have been multi core since the early days of TNT, theres no need to hope for a multi core GPU, they are already here.

Isn't it odd ATI and Nvidia doing multiple GPU's... 3dfx went under because of it...

3dfx went under for lack of support and lack of ability to get their product out, their multi GPU market was their bread and butter, it was their inability to get their product on the shelves and support for 32bit 2d that hit them hard.

512bit memory achritecture is not in our near future, we still haven't tapped into the true potential of 256, i doubt the PCB makers would be happy if either vendors tried to push a 512 architecture down their throats.

Memory can get so much faster but it's going rather slow right now and to keep up with Nvidia and ATI 6 month rehash and whole new generation every year...

memory is getting extremely fast, especially with the debut of GDDR4, which is coming to us with the X1950XTX, memory bandwidth has shown to be rarely ever been the bottle neck of GPU's, with a better memory controller (ATi's ring bus for expample) we might see some exponential increases in performance, but the story with most cards is the GPU's throughput trying to catch up, I honestly doubt anyone is trying to come out with a 512 bus anytime soon.
 
The whole Quad thing is retarded to the boot in my opinion. I would never even consider crossfire so Iam quite glad if Ati does not follow Nvidias path. And I think there is quite enough reasons for sticking with one card, no need to list em.

I mean where is the innovation in slapping more same cards together to get performance. Single card is the only way for me.
 
What I dont understand is all these people saying how stupid Quad is are running normal resolutions. Try running oblivion on a 3007 and some candy and see how your tune would change.

Now on the other hand spending the cash isnt everyones cup of tea... but it IS for some, and having the ablity to do it is awesome. Everyone wants innovation but some dont have the imagination to see it for what it is... an evolution.
 
Trimlock said:
3dfx went under for lack of support and lack of ability to get their product out, their multi GPU market was their bread and butter, it was their inability to get their product on the shelves and support for 32bit 2d that hit them hard.

512bit memory achritecture is not in our near future, we still haven't tapped into the true potential of 256, i doubt the PCB makers would be happy if either vendors tried to push a 512 architecture down their throats.

memory is getting extremely fast, especially with the debut of GDDR4, which is coming to us with the X1950XTX, memory bandwidth has shown to be rarely ever been the bottle neck of GPU's, with a better memory controller (ATi's ring bus for expample) we might see some exponential increases in performance, but the story with most cards is the GPU's throughput trying to catch up, I honestly doubt anyone is trying to come out with a 512 bus anytime soon.



3dfx had great support... It was their lack of business decisions to not use DDR and not use 4 pixel per pipes like geforce did... That really hurt them big time... Not to mention 1 geforce ddr can match dual VSA 100 chips which cost more to make... Glide was coming out... Opengl and DirectX was the standard... It was a lot of things... You get the idea...

That's what I'm saying 512bit memory controllers is still few years away... GDDR4 is not all that much faster than GDDR3... Remember ATI and NVIDIA is trying to double their performance every generation... At least that's what they'be been doing for few years...

Radeon 8500--------->9700pro--------->x800---------->x1800---------->1900 nothing more than a rehash with higher clocks to me....

Last generation slowed down quite a bit and nothing more than a clock increase and a little better memory.... Hence they're pushing more dual chip configs... They're hitting that wall again with current technology....
 
I don't think many are wondering how they will compete. Well, first impressions are built by first releasers. The idea of 4 gpus, craptastic AF and jaggies in Oblivion is laughable at best. I honestly could have all the money in the world and Quad SLI would not interest me one bit.

Now I guess if i was loaded, I'd wait until either Nv gets with the times or ATi makes a quad crossfire that might not give a 1kw pcp&c powersupply all it can handle. You know, in that case I might be loaded but it would seem silly to have two ACs in my room just so my balls wouldn't stick to my leg. Gaming isn't that fun.
 
hardwarezone using an older CPU, and 1 gig of craptastic ram? nice review...

3dfx had great support... It was their lack of business decisions to not use DDR and not use 4 pixel per pipes like geforce did... That really hurt them big time... Not to mention 1 geforce ddr can match dual VSA 100 chips which cost more to make... Glide was coming out... Opengl and DirectX was the standard... It was a lot of things... You get the idea...

3dfx had already lost the battle and the war by the time the first geforceDDR rendition got together, the TNT2 is what really did it in for them, they had lacked the key component of supporting 32bit 2d, while Nvidia touted it all day long, the fact they never came out with a 4pipe design card with DDR was a little late, if they came out with that while the first Geforce did not, they could still be in the game, they still didn't get that far.

I can definetly say if 3dfx was still around, glide would still be around as well, they loved it and loved giving developers support for it so they could put it into their games, i really miss glide support.

That's what I'm saying 512bit memory controllers is still few years away... GDDR4 is not all that much faster than GDDR3... Remember ATI and NVIDIA is trying to double their performance every generation... At least that's what they'be been doing for few years...

Radeon 8500--------->9700pro--------->x800---------->x1800---------->1900 nothing more than a rehash with higher clocks to me....

Last generation slowed down quite a bit and nothing more than a clock increase and a little better memory.... Hence they're pushing more dual chip configs... They're hitting that wall again with current technology....

DDR4 on its first rendition into the market shows at least a 10% increase, thats more then DDR2 ever showed for video cards, i'd say DDR4 is going to become godly for VC's, much like DDR3 was.

8500 is no where near anything like hte R3XX chips, nor is the X1800 or X1900 anything remotely close the R4XX chips, hell even the X1800 and X1900 have quite a few differences, definetly not rehashes and nothing more hten just a few increased clocks.

and the only wall i see GPU's hitting is the power requirements, adding more features and engines is going to cost them more space, having to put more transistors is always going to require more power, and cause a hotter chip, hopefully soon they'll create a designe that doesn't require a billion seperate engine to push that data and have a smaller die.
 
What I'd REALLY love to see is the "jetway" system partitioning done with GPU's.

One physical box with a dual core = 2 cores. Jetway has systems that supposedly allow two users to work at the same time on the same physical box as a local terminal. Hook up 2 KB/mice/monitors and you are set.

What would REALLY be sweet would be to have a dual gpu card that run 2 3d apps at the same time and have display be completely separate. I'd but that in a heartbeat.
 
By the way, I think this is more like the third or fourth Nvidia card with multi-GPU technology on it. Gigabyte had a 6800gt with dual GPUs and I think ASUS had a 7800gtx with multi-GPUs IIRC (not 100% sure about the ASUS one).
 
Trimlock said:
3dfx had already lost the battle and the war by the time the first geforceDDR rendition got together, the TNT2 is what really did it in for them, they had lacked the key component of supporting 32bit 2d, while Nvidia touted it all day long, the fact they never came out with a 4pipe design card with DDR was a little late, if they came out with that while the first Geforce did not, they could still be in the game, they still didn't get that far.

I can definetly say if 3dfx was still around, glide would still be around as well, they loved it and loved giving developers support for it so they could put it into their games, i really miss glide support.

DDR4 on its first rendition into the market shows at least a 10% increase, thats more then DDR2 ever showed for video cards, i'd say DDR4 is going to become godly for VC's, much like DDR3 was.

8500 is no where near anything like hte R3XX chips, nor is the X1800 or X1900 anything remotely close the R4XX chips, hell even the X1800 and X1900 have quite a few differences, definetly not rehashes and nothing more hten just a few increased clocks.

and the only wall i see GPU's hitting is the power requirements, adding more features and engines is going to cost them more space, having to put more transistors is always going to require more power, and cause a hotter chip, hopefully soon they'll create a designe that doesn't require a billion seperate engine to push that data and have a smaller die.

It really wasn't about 2D with 3dfx... 3dfx and ATI cards had better 2D than TNT... TNT2 had 32 bit color but to actually use it games was major drop in performance during that time... Most TNT2 guys used 640x480 with 32 bit while voodoo was using 16 bit higher quality 1024x768... Looked almost good as 32bit IMO... Lot of people bought into that techno garbage but that's how it is sometimes... Fact was voodoo 3 was much smoother card than TNT2... As for voodoo 5 well they should have used what was available to them... 4 pipes with DDR... If they did they would have destroyed geforce1 and 2 and even compete today's 8 pipe cards... The fact that they used 2 pipes and didn't try to sell their chips to other companies like EVGA, Leadek, etc.... and used multi GPU in 1 PCB is what hurt them...


From SDR to DDR increased 100% fold... And while DDR 2, 3, and 4 is just clock increases but a major increase at that... But not so recently since these cards have over 30 GB/s memory bandwidth...

Actually all radeon chips share the same technology... It's just a build up of the same chip... Constantly enhancing it... Adding more pipes/clock... Adding Pixel shaders, vertex shaders, AA.. etc...

x1800 is very similar 1900 series in fact we can see this when you compare 1800xt against 1900gt... They are simliarly clocked... And give off similar performance...

When I say cards hit a wall... I mean that you can't increase performane other than clock inreases and 10-20% faster memory.... GPU's are already what at 90nm... 65nm are not ready yet... They can't add more pipes because the die isn't capable yet...

http://www.theinquirer.net/default.aspx?article=33428

Seriously I wouldn't mind a gaming card to have their own external power supply... This would be great idea even for people who don't have power supplies that can handle them... I would like to see this in the near future...
 
It really wasn't about 2D with 3dfx... 3dfx and ATI cards had better 2D than TNT... TNT2 had 32 bit color but to actually use it games was major drop in performance during that time... Most TNT2 guys used 640x480 with 32 bit while voodoo was using 16 bit higher quality 1024x768... Looked almost good as 32bit IMO... Lot of people bought into that techno garbage but that's how it is sometimes... Fact was voodoo 3 was much smoother card than TNT2... As for voodoo 5 well they should have used what was available to them... 4 pipes with DDR... If they did they would have destroyed geforce1 and 2 and even compete today's 8 pipe cards... The fact that they used 2 pipes and didn't try to sell their chips to other companies like EVGA, Leadek, etc.... and used multi GPU in 1 PCB is what hurt them...

doesn't matter if it gave a huge performance drop, the fact that they had it, everyone wanted it, Nv bragged about it for a long time, came out with it, and the world flipped. Releasing the voodoo5 was a mistake, they should of re-released some of their low end product first, the voodoo5 had no chance of making it out to the public in mass and never did, it was a sought after item, better then the competition? no but it was a card alot wanted (namely name recognition).

From SDR to DDR increased 100% fold... And while DDR 2, 3, and 4 is just clock increases but a major increase at that... But not so recently since these cards have over 30 GB/s memory bandwidth...

like i said, throughput trying to keep up with the memory

Actually all radeon chips share the same technology... It's just a build up of the same chip... Constantly enhancing it... Adding more pipes/clock... Adding Pixel shaders, vertex shaders, AA.. etc...

no they aren't, the 8500 was completely in house created, the R300 chip was developed by a company other then ati, the X800 line of GPU's were extremely similar to the R300's in the fact it was pretty much the same die with a little extra support (same thing with the G70/71 and the NV40), the X1K family is a completely different GPU, built from the ground up much like the XENOS was, these GPU's are completely different and only share similiarities but are not just further revisions of the previous gen.

When I say cards hit a wall... I mean that you can't increase performane other than clock inreases and 10-20% faster memory.... GPU's are already what at 90nm... 65nm are not ready yet... They can't add more pipes because the die isn't capable yet...

you can add as many as you want, you'll get less and less GPU's capable of actually using all of these and you have to worry about getting power to these but you aren't limited in the size, its just not practical.

Seriously I wouldn't mind a gaming card to have their own external power supply... This would be great idea even for people who don't have power supplies that can handle them... I would like to see this in the near future...

external/internal they are already here, and i like the idea too, it should be the norm with the next gen coming from both companies, this is something i'd want with the X1900XT i have.
 
no they aren't, the 8500 was completely in house created, the R300 chip was developed by a company other then ati, the X800 line of GPU's were extremely similar to the R300's in the fact it was pretty much the same die with a little extra support (same thing with the G70/71 and the NV40), the X1K family is a completely different GPU, built from the ground up much like the XENOS was, these GPU's are completely different and only share similiarities but are not just further revisions of the previous gen.

That's not entirely true... it's true ARTX stepped in and made the r3xx chips but it's still the same tech... plus pixel, vertex shaders... 8 pipes, 256bit memory controller.. and adde tri-aniso.... they just did a better job at executing...
 
The main reason Quad anything doesn't really intrest me is simply the law of diminishing return.

Such a huge money investment and the more you add the less you get for your $1.00
 
O.T. 3dfx went under more due to idiotic management decisions than technical flubs or anything else.. The cutting out of 3rd party boardmakers and purchasing of STB was a couple of the most idiotic things they ever did, followed closely by not open sourcing glide til it was already too late for the api..

B.O.T. I can see Ati, err Amd, err whoever, putting out something similar.. Quad SLI is not exactly a gimmick, but it really is a product meant more for the marketing dept than the actual consumer.. But, it lets them say " We be the baddest pixel pushen mofos in the land, fo' shizzle"......And that seems to sell video cards.. Even if is only the hardcore with plenty of disposable income that would ever really ever buy quad sli, or for that matter, topend Crossfire/Sli setups ..

What will they call it I wonder .. Crossfire Xtreme, Ultra Fire, Uber Crossfire.. I will see I suppose..
 
Chris_B said:
Personally i think this mentality of "omg slap more gpu's together for more fps!!!" is getting retarded. At the most regular sli or crossfire is more than enough for most people, then nvidia come out with their *cough* 2 card *cough* 4 gpu solution How long before we see 6 or 8 gpu's slapped together? Just seems to be getting out of hand.

quad SLI, is not a mainstream product, its most for enthusiast,

just like saying ferraris are stupid, i'm happy with my kia

totally different market.
 
Back
Top