Slashdot Mirror
Nvidia Reintroduces SLI with GeForce 6800 Series
An anonymous reader writes "It's 1998 all over again gamers. A major release from ID software, and an expensive hotrod video card all in one year. However, rather than Quake and the Voodoo2 SLI, it's Doom3 and Nvidia SLI.
Hardware Analysis has the scoop, 'Exact performance figures are not yet available, but Nvidia's SLI concept has already been shown behind closed doors by one of the companies working with Nvidia on the SLI implementation. On early driver revisions which only offered non-optimized dynamic load-balancing algorithms their SLI configuration performed 77% faster than a single graphics card. However Nvidia has told us that prospective performance numbers should show a performance increase closer to 90% over that of a single graphics card. There are a few things that need to be taken into account however when you're considering buying an SLI configuration. First off you'll need a workstation motherboard featuring two PCI-E-x16 slots which will also use the more expensive Intel Xeon processors. Secondly you'll need two identical, same brand and type, PCI-E GeForce 6800 graphics cards.'"
All great news.. but WHEN can I find it available in the stores, that would be NEWS.
// instant - "I for one welcome our new Decaff Coffee-Flavoured-Coffee Overlords"
There goes my savings again!
Its sad that my first born has to go..
But perversly exhilarating to hold an SLI configuration in my hands instead..
Rapid Nirvana
These cards are expensive enough, now they are suggesting we buy 2!?
I guess if you have a lot of money and want to play with a (marginal) advantage, an SLI setup is for you.
As for myself, I am a poor college student not even able to afford 1 of these cards. A situation I think is similar to a lot of other geeks/gamers.
Which begs the question, who is this aimed at?
"Here's a spoiler: You're will die alone."-Triumph the Insult Comic Dog
What the hell does SLI mean? And why does anyone care? I.e., what are the real world (pc gaming) results of paying more money than I can afford to use this technology?
Am I the only one to find it hilarious that at the top of the page there was a Flash ad for an ATI Radeon X800?
"Get a bicycle. You will not regret it, if you live." - Mark Twain, "Taming the Bicycle"
... till we have multi-core and/or multi-GPU consumer cards. (they're already available at the high-end)
Questionmark.
Belief is the currency of delusion.
sweet... so now i can get the same performance using my 3d Glasses as all you 2D slacker for only 2 or 3 times the price... wicked!!
(now = sometime in the future)
Break out my old Voodoo2 cards again... SLI? Sure... My voodoo2 cards in series from my Radeon x800 :-) Who says you have to have a 6800? ;-)
like 3dfx they bought?
maybe they shouldn't have.. sure they probably had some great people and so on but ultimately "it didn't work out".
"hey, we can't keep up! let's just use brute force on increasing our cards capabilities!!! that's cheap and economical in the long run keeping our company afloat, right? right??"
world was created 5 seconds before this post as it is.
Can you hook up 4 monitors to this badass configuration?
So THAT'S how we can run Longhorn! It makes sense!
Oh wait...
So, One card that requires a 400 Watt power supply + Another card that requires a 400 Watt power supply = The need for an 800 Watt power supply?!
It's a bit presumptuous to assume that when these SLI cards come out, the only motherboards supporting multiple PCI-E x16 slots will be Intel Xeon based. As far as I knew, AMD were planning on doing 939 based motherboards with multiple PCI-E.
At any rate, doesn't this sort of make the whole Alienware Video-Array seem like a bust?
/richboyon Anyone who says that this kind of setup isn't necessary simply cannot afford the necessary gear. I am looking forward to the envy and hate from all the punk kids playing with their mom's computer. Truly the Ferrari concept brought to the desktop. /richboyoff
Now, what is a realistic price for a system such as this?
Why don't they make a graphics card with two GPU and double memory size ? Or wouldn't fit on of these buggers into a computer case ? Yes they exploit the dual PCIex busses, but it doubt that they really use the would bandwith.
Am I the only person who thinks that holding the two together with a non-flexible medium and is held on only with solder is a bit dangerous? Not that the solder would break, but when it is removed, it could be a bit tricky. Perhaps a cable on there would be safer.
Other than that the only problem I can see is that you need about AU$2000 worth of video card, and at least AU$1000 worth of Xeon to use it. Maybe for engineers and artists, but will the average person have any use for it? I don't feel that an extra AU$3000 is worth it for the extra frame rate in games.
For the pros though it would be very good though.
The lower-than-100% increase reflects the fact the cards aren't working together fully. As they said, it's still early days, and expect to get that figure to nearer 90%.
It's easy to understand. They're getting their asses kicked routinely by ATI, so they have to resort to 3DFX-like measures to try to get some real performance happening. What a shame.
If you're going to criticise, double-check your own figures. Adding another card is 100% more for both GPU and expense, not 200% (that would be three cards).
How does this stack up against Alienwares ALX dual graphics card system. I remember reading an article where the Alienware guys bashed the SLI method. Theirs, each card renders half the screen, either top or bottom, not every other line.
I want 2D games back.
And I'm also wondering how the heat is going to be transferred away from the cards. It looks like you need some serious cooling setup to keep those two babies running.
The best weapon of a dictatorship is secrecy, but the best weapon of a democracy should be the weapon of openness.
um... maybe you mean 100% more GPU
or 200 % the gpu, 177% the power
You mean:
100% more GPU.
100% more expense on said GPU.
77% more performance???
Right?
I don't know where you got 77, nvidia claims their target is 90% but I'm going to guess the real numbers for general use are a bit below that.
Call me when they put two GPUs on one card... Or even better, when they put two cores on one chip. Soon enough motherboard will be an add-on to graphic card.
Plus, many people were upset about power and cooling requirements. This monster would occupy FOUR slots and require, what, a 600W PSU? (ok, just kidding, "only" 460W should be enough)
Who needs a new car? A rig like this looks like something worth building. Time to start looking for an interesting new case and a place to put another PC.
I think you may mean 100% more GPU, and 100% more expense. 200% would be for 3 cards.
77% more performance because the drivers are not yet optimised, and because I'd assume that the tests were done on the same machine.
Not all of the processing is done on the video card, what makes you think that another one will take all the load off the CPU?
100% more GPU and expense. 77% more performance. It's a better deal than going from the slightly-obsolete products to the first tier, anyway...
Free Java games for your phone: Tontie, Sokoban
I picked up a Voodoo 2 card way back when for the incredibly high price of $300 (which was a ton close to ten years ago with the money I was making). A couple years later, I picked up my second Voodoo 2 for $30.
Think of it as an inexpensive way to nearly double your video card's performance at a fairly cheap price when others are upgrading to the new version of the card that is only 40-50% faster (unlike the SLI mode which is rumored to be 75-90% faster).
The tricky part will be that you have to have a motherboard to support it, which for now will only be the ones made for high-end workstations.
It's a pretty simple case of diminishing returns. If there are now 2 cpus in charge of doing rendering, they have to spend some of their power cooperating and communicating rather than just crunching numbers all of the time.
Of course that is a terrible over simplification. There are cases in which 2 cpus are actually slower than one, notably SMP P1 chips that had the L2 cache on the motherboard.
+++ ATH0 +++
1: Resort to idiotic 3DFX-like measures to get high performance
Note: A 77% increase in gaming performance isn't "high performance". Considering that the 6800 is ALREADY a massive leap forward over it's predecessor, it's INSANE PERFORMANCE!
How would something like 1600x1200 with maxed FSAA and maxed AF, while never dropping below 60fps, grab you by the short and curlies?
2: Watch company slowly die.
Nobody's suggesting that everyone and their brother run out and get SLI'd GeForces on a Xeon platform. (Those already spending 4-5000 dollars on such a platform aren't necessarily going to shrink from an additional $4-500, especially if it nearly doubles video performance.)
This is going to probably be limited to those who'd normally use Quadro cards (productivity) and the elite few with more money than sense.
Not that everyone won't WANT one...
Chas - The one, the only.
THANK GOD!!!
Your bank loan is ready. This is beyond pointless. At least it was affordable with the Voodoo2s.
One of the 187.
great nvidia, 1 Kw PSUs now needed no doubt.....
and four slot spaces? eek. I'll wait for the ATi MAXX solution (you know they'll go one better), that way we dont need dual PSUs (the nvidia requirements are stupid, seperate power rails per molex connector, not allowed to share the spare connector on each rail? 8 connectors in a sense.. jeez) or liquid nitrogen cooling..... (id never ever try a dual 6800 ultra with the default sinks, eurgh).
If you ask me, 2x 6800 GTs or so are cheaper, almost as fast, less requirements and less space. Although new mobo would still be needed for dual PCI-e.
Do a MAXX or Voodoo5 for crying out (hell, put the second card on a daughterboard and using the same slot) just don't do a Volari.........
PS, Alienware can smeg off. rip off sods. It isn't $5000 for a new mobo and 2 cards in a system that would cost 1000 or so and be kickass...
All great news.. but WHEN can I find it available in the stores, that would be NEWS.
Dude, this is an NDA leak! If you're trying to imply nVidia is peddling vaporware, well, you might be right, but in this case they're actually not the ones doing the peddling, because their SLI setup is still under NDA.
I'm in two minds about this. On the one hand, only a tiny number of games require anything like this sort of horsepower. But on the other hand, These sorts of expensive, high-end systems will be commonplace in 2-3 years time. I bought a GeForce FX 5900 when it was the top-end card, and now it's (kind-of) obsolete!
Cress, cress, lovely lovely cress
The proliferation of aimbots and wallhackers will still mean you just look better getting pwn3d.
Do not try to read the dupe, thats impossible. Instead, only try to realize the truth
What truth?
There is no dupe
Buy them here.
Life is the leading cause of death in America.
John Carmack said about a year and a half ago that Doom 3 would run 'well' on a top-end system of that time- which was a 3.06 GHz P4 equipped with a Radeon 9700 Pro. What's frightening/upsetting is that this SLI setup really isn't coming into play to satisfy the games of today like Doom 3- it's coming into play for the games of next year and the year after. It's just a little off-putting that in order to play the newest games you need a SET of graphics cards with those kind of power and space requirements.
...priorities. If gaming is your life (or if you're a working man with a gaming fix), two of these aren't that "extreme". People easily spend 10k$+ more on a car than a car that'd get them from A to B just as safely and easily, just for style and more luxury.
...and that was the 3rd webshop I had to go to in order to actually find one of those - most now have some legacy 17 and 19" CRTs and the rest LCDs, which go no further than 1600x1200 (even at 21") and don't need an SLI solution.
If gaming is what you do a considerable number of hours of your life, why not? Even as a student, it'd be some weekends without being completely wasted (and maybe work an hour or two as a weekend extra), and you'd have it.
All that being said, from what I saw with the last cards it looked to me like GPU speed was starting to go beyond what conventional monitors and CPUs could do. And those really huge monitors are usually far more expensive than the GFX cards, even two of them.
2xGF6800 = 10000 NOK
Sony 21" that can do 2048 x 1536/86 Hz = 14000 NOK
Personally, I'll probably stick to GF4600 until hell freezes over, I just don't manage to get hyped up on the FPS games anymore. I'd rather go with a HDTV + HD-DVDs, should they ever appear...
Kjella
Live today, because you never know what tomorrow brings
> Which begs the question, who is this aimed at?
I recently learned this here, so please don't take this as a criticism.
The phrase "begs the question" doesn't mean what you think it means. It does not mean, "this leads to the question."
Rather, it is a term used in logic to indicate a fallacy in which the question or statement itself tries to prove its truth by asserting its own truth. This is commonly known as circular reasoning. More here.
I agree with you about wondering who the product is aimed at, though.
The target reader is assumed to know what SLI is already. You are therefore not meant to be here . . . so go away!
Whoa...with one GPU power requirement being a 500W power supply. Where the Hell am I going to get a power supply to run two of these beasts. Soon my PC is going to draw more than the beer fridge!
This is the mobo design Alienware came up with, right? My understanding is that you can use ANY two video cards that are the same and are PCI-X. You could just as well do two ATI cards. Who submitted this? Nvidia marketing? :-)
Have a Happy.
Please register or login. There are 9 registered and 3599 anonymous users currently online. Current bandwidth usage: 872.66 kbit/s
:D
How many of you are there hitting refresh just to see the hit counter go up?
Sometimes I wish I was a plumber, then I'd know how to deal with other people's shit.
SLI stands for Scan Line Interleave.
Patrick Doyle
I mod down every jackass who puts his moderation policy in his sig. Oh, wait a sec....
I seem to remember that one of these cards took up 2 slots, and needed a third just for good air flow. How much space are these going to take up? Also, just one of these bad boys needed something like 400-500W of power. What kind of power supply is needed for 2???
When a single 6800 card requires a 480watt power supply and two dedicated power lines, what would the power requirements be for two of these cards in the same computer system?
...means you have maximum power draw from everything at once (and maxed out on "empty" slots too). Your 160W PSU might have been able to run the system at idle, but at max load it might fail.
Hard disks are not a major draw of current anyway, I checked Seagate's website and they draw 13-14W at max each. It's the 100W+ CPUs + 100W+ GPUs + cooling systems that make up the most part. Each PCI slot also adds a lot to the requirement, since they can each draw a lot (even if they fairly rarely do, depends on type of card).
Kjella
Live today, because you never know what tomorrow brings
Why would they design something like this and force it to use a Xeon?
For starters, the Xeon is still stuck at a 533MHz FSB, limiting its performance. Add in the fact that they're ridiculously overpriced & most games show little to no performance improvement when running on an SMP system. A single P4 or Athlon64 will stomp the Xeon in almost all gaming situations.
Of course, with this tech a ways away & there not really being any PCI-E motherboards on the market now that Intel's recalled them all, I guess they're betting on high-end enthusiast boards to ship with the second x16 slot by the time this thing is actually ready for market...
Really, the biggest application for this kinda power that I can forsee would be game developers who want to see how well their games scale for next-gen video hardware...
my sig's at the bottom of the page.
Quake 3 came out in 99, Quake 2 in 97.
Hanging in my closet is a "souvenier" from my last adventure with SLI: A Quantum3D Obsidian t-shirt. In my eagerness to own the latest and greatest graphics card I paid 600 bucks up front to preorder this card which was developed by a spin-off from 3DFx. The card shipped 6 weeks late, suffered from overheating since it crammed the components from 2 cards into a single PCI slot, and was soon equaled in performance by a simple pair of Voodoo 2 cards in adjacent slots. I expect a similar fate for this monstrosity since the GeForce 6800 pulls what - 75 watts ? I assume a 500 watt power supply will be required. Thanks, but no thanks.
-The Mad Duke
Just doens't make much sense to me. If there's a game my Boxx FX53 + X800 won't play well, then it's probably not worth playing.
Every serious gamer knows that 86Hz is unacceptable. True gamers know: CRT > LCD / PLASMA. Until you can find me a plasma that can refresh at 125Hz or greater, I'll stick with my 80lb. CRT.
Any gamer extreme enough to buy two of these cards plus the requisite hardware should be smart enough to know that a flat panel is a waste of money for games. Then again, they are gamers...
how well does it overclock? :-)
More seriously, SLI is good because it protects your investment, but only if it really gives around 2x the throughput. Consider this: If you buy a single 6800 now and in a coupla years it cant keep up any more it gives you the option of buying another when they would be relatively cheap. I guess the only problem will be trying to buy *exactly* the same card twice with a 2 year gap inbetween.
The G5s come with PCI-X slots, and will prolly run Doom 3.
:)
Usually the cost of G5s are a bit too steep, but compared to Xeon they are an absolute steal!
And you get a cool machine/OS to boot!
Great! But...
I already find limitations in the software in getting realistic images far below this. What we need right now isn't more fps, and it's not resolution either. We need more triangles. What good is humongous resolution, if every tree is shown as a couple of crossed bitmaps? I will not invest in better hardware until the software is designed to be more realistic.
OTOH, for those who need professional workstations, e.g CAD users, they can do well with resolution without 3d performance. Why should you need 100 fps to see a revolving crankshaft?
The question remains as to whether or not it will be worth having to fork out the extra dollars just for a bit of an extra performance bonus, and the fact that you'll be able to just say that you have two video cards running in SLI in your machine. For the average consumer, it isn't really going to make much of a difference. For the bleeding edge gamer... we'll just have to wait and see.
-
"I may have invented it, but Bill made it famous." - David Bradley, inventor of Ctrl-Alt-Del
If you don't remember what it is, just look at the faq.
I feel that this dual card thing will not be as short lived as the old 3dfx SLI. I mean, it wasn't possible to use 2 AGP cards because we lacked the second slot, but with PCI-E, the problem is gone. Remember that all the Voodoo2 had the SLI plug ? I bet that all the next gen cards will have a dual mode plug (it's already the case with the new GeForce).
The next step is to allow this kind of thing with non-identical cards. It would be nice to be able to keep your old card even after you've bought a brand new one. But it seems that synchronization is a bit of a problem.
The article calls this 'SLI' 'Scalable Linked Architecture'.
Indeed, it uses a top/bottom 50/50 split for rendering rather than per-line interleaving.
Ripping an new rectum in the fabric of spacetime.
Does using 2 video cards really help when we can't use what's alread there?
Ummm.. yes?!
Video cards use bus bandwidth. If we, as you point out, have bus bandwidth to spare, then one solution to using it up would be to put in more cards.
Come on, this isn't exactly rocket science at this'a'here news for nerds site!
This thing will work only if both GPU have access to each other videomemory. All the other attempts to use paralell GPU failed exactly for that reason(I'm talking about moredn cards here) . It's not clear from the article if it is the case. If it's just two GPU working in paralell, all the old problems of paralell GPU raise - GPU can not divide vertex processing load, only pixels, texture should be loaded into each card sepately, cutting memory bandwidth twice, synchronization etc. However if both GPU share videomemory all this problems go away.
I'm not a native english speaker, but that does not sound right ^^;
Moderators: Don't agree? pray tell why.
According to the article...
In essence the screen is divided vertically in two parts; one graphics card renders the upper section and the second graphics card renders the lower section.
Now, as we all know, old school SLI mode meant that that each card drew every other line. This one, however, looks to be more like Alienware's implimentation, with each card drawing half the screen. nVidia needs to find a terminology and stick with it.
There is no mod option "-1: Disagree" for a reason. "Overrated" is not an acceptable substitute. Post something instead.
No point in complaining. Let the folk rich enough (stupid enough?) to afford it, buy it. Either it just won't take off (in which case you've saved yourself a load of cash) or it'll go great, the price will drop, the bugs will be ironed out and you'll get it at a price you can afford. What is there to complain about?
---
We spoke for about a half an hour. I don't recall a thing we said. - Colorblind James Experience
200% more GPU.
200% more expense on said GPU.
And 200% more problems calculating percentages!
Ironically, the word ironically is often used incorrectly.
and obviously, two cards might be 100% louder than 1 ...
me won't buy that !
That's all I am seeing here. You don't need to use the Ultra in your configuration of this. The article even states you can use a single slot GT, which would be greater than a single Ultra and cost you 200 dollars more for a great performance boost. Or you could even use basic 6800 cards which are under 300 dollars.
This is going to be great when it matures, and is one of the huge advantages to PCI-Express when that becomes the standard on future motherboards over AGP. Yes, I know Intel is making motherboards with this, but who the hell wants to pay all that money for such a small jump?
Since people seem to be lost on the nvidia cards, here goes a run down of what they are releasing and the price area:
300$ - nvidia 6800
400$ - nvidia 6800gt
500$ - nvidia 6800 Ultra
600$+ - nvidia 6850/6800 Ultra Extreme
The 6800 and GT are single slot cards with a signle Molex connector. Those can be used in the SLI configuration as well. Get the facts straight before you post flamebait and troll.
That's scary.
Not only would you have a kickass system to play Doom3 on, but you'll be able to heat your entire house with it.
Imagine a beowulf cluster of video cards in your machine.
IT COULD HAPPEN
make this GPU nonsense moot by this time next year. How? Their new cell-based PC platform.
LinuxInsider has a couple of recent op/ed pieces about this technology. Not sure I agree with the timeline suggested, but at 10X the speed of current processors by next year or so, Moore's may need a revision.
An AMD 64 notebook is moving into my place soon anyway and so will watch the new tech to stir up the pot in relative comfort.
Everything in the Universe sucks: It's the law!
So, lemme get this straight-in order to get a 77% speed increase, I'm going to have to blow hundreds on a second card ($400), xeon processor, motherboard, memory, and a damned good cooling system so it all doesn't melt and I don't go deaf? Wouldn't it make more sense to buy a decent card now, and wait two years for them to put out the single GPU card that does the same performance for $200? Unless you're really worried about dropping under 100 frames, or you have a lot of high end rendering to do, I can't imagine this really being worth it. At least with the Voodoo 2 SLI system you could buy a second card without having to invest in a huge honking system that makes a dual G5 look cheap.
You could wait 3-4 weeks and pick up a SLI capable card that is four times as fast as a dual card setup and save $900 doing so.
This is interesting, not because of added performance, but because of the ability to add more than two monitors for a gaming setup. Matrox Parhelia is the only current card (that I know of) that can do tree monitors, and its way too expensive and doesn't perform that well either...
I just hope this SLI technology gets into some cheap model cards too, buying two cheap cards could offer good performance and three (or four) monitor setup with an affordable price.
The above post makes almost no sense, don't go modding it up. I'm not even going to waste time going into the specifics.
The Vodoo 5 5500 had two processors and the Voodoo 5 6000 had four. If they would have come to market fast enough and not been super expensive for the time, it really would been something.
[RIAA] says its concern is artists. That's true, in just the sense that a cattle rancher is concerned about its cattle.
I wouldn't.
What I would like to see is 30 fps when I throw around 2-5 million triangles on the screen. High resolution (half a meter in rw) GIS data is a pain to render fast. Sometimes you just have no choice, LOD algorithms are not always acceptable.
Current top of the line radeons are pushing $600 around here, I can only imagine what this thing will cost.
Think I'll stick with my Xbox.
You want to play Doom3. You could either spend a small fortune on a workstation computer with dual GPUs or you could spend about 150 bucks on an Xbox, probably even less by the time the game is actually released. Mmmm... what to do what to do.
If someone says he and his monkey have nothing to hide, they almost certainly do.
ATI released a card with multiple GPUs on it a few years ago, putting a pair of Rage 128 chips onto a single card. It provided at best a marginal performance increase, but was still a neat idea at the time.
More info here.
I made a post however many years ago about 3dfx and how it is retarded to do SLI style setups. Of course back in the day you needed to spend fuck knows how much on Dual Voodoo cards and also a proper graphics card to handle 95% of what you did.
My opinions still hold true, i'll wait a month or two for the next GPU and get a single card version.
"There are a few things that need to be taken into account however when you're considering buying an SLI configuration. First off you'll need a workstation motherboard featuring two PCI-E-x16 slots which will also use the more expensive Intel Xeon processors. Secondly you'll need two identical, same brand and type, PCI-E GeForce 6800 graphics cards.'"
...and they know you're going to rush out to buy them, as the almighty Carmack wills it! You want your precious Doom3, don't you, geek? Here's your next hoop! Jump! Jump!
Although the Parhelia wasn't a gaming success, with the masses tending toward FPS count above all else, the Parhelia is a great card. I'm guessing features such as 'surround gaming' are not available on the Nvidia fan heater. Given that matrox have sat quietly since the Parhelia's release, only tweaking it here and there, i'm sure there working on something. I'm holding out for Matroxs answer. That, and my room is warm enough.
has a good look at this whole thing for all of the reposts im reading about the power and the space and the well im sure you know what i mean.
0 40628/index.html
heres a link http://www20.graphics.tomshardware.com/graphic/20
"On early driver revisions which only offered non-optimized dynamic load-balancing algorithms their SLI configuration performed 77% faster than a single graphics card. However Nvidia has told us that prospective performance numbers should show a performance increase closer to 90% over that of a single graphics card."
Certainly the slashdot crowd will have noticed that this has nothing to do with SLI, but i don't think it's been mentionned here that the technology is actually PGC. Even though i remember it clearly, it didn't seem to get much press at the time. I found a story about it but dear Tom seemed to be a non believer.
Ok, we haven't seen it yet, but with Alienware and nVidia backing it, i'd be surprise if it didn't show up, one way or another.
Let us forget for a moment that I could have set my preferences up to ignore inane comments like this; let us also forget for a moment that I could be a good little boy and answer this question.
Now, I think I speak for a great many readers on here when I say this: RTFA!
However, I have to wonder, if you can't be bothered to read something that is already provided, why should we think you will read something that is provided again, here? Nevermind the fact that this VERY FUCKING QUESTION has already been answerd no less than THREE FUCKING TIMES in this thread!!!
Please, just get out of the fucking gene pool. Go find yourself a cliff with some large, sharp rocks at the bottom and jump off. I feel dumber just by sharing the same SPECIES as you, but I think that you killing yourself will make me feel better, to say nothing of somewhat redeeming yourself in the eyes of all present.
There is no mod option "-1: Disagree" for a reason. "Overrated" is not an acceptable substitute. Post something instead.
here.
as AlienWare PC already introduced their plan of a simple dual PCI express mother board to support two video card run in parallel, further more, it can support ANY two video cards, instead of identical ones. FX6800ultra demands 480W PSU to run, what does a dual card configuration demands? 600W PSU? with the super hot xenon CPU in the case, better put the machine in the kitchen and use it as your stoven...
My new
Here is a much more detailed article about this
HardOCP Article
kawai
Shame on me for not RTFA :(.
Still considering the price, 77% does not jibe too well with me right now.
"Here's a spoiler: You're will die alone."-Triumph the Insult Comic Dog
"First off you'll need a workstation motherboard featuring two PCI-E-x16 slots which will also use the more expensive Intel Xeon processors. Secondly you'll need two identical, same brand and type, PCI-E GeForce 6800 graphics cards"
They say this like I need a resaon to spend too much money on computer hardware.
TruePunk | Games
my Boxx FX53 + X800
Their target market is apparently "you" - you're just in the wrong place in your cycle. Right now, you're in the sour grapes phase, denying the possibility that anyone could want a better computer than yours (they already do). Soon you'll be in the lust phase, then you'll be in the "MUST BUY SHINEY THING! PLEASE TAKE CREDIT CARD!" phase.
I remember a time when it was unimaginable who might need a 386.
Let's not stir that bag of worms...
Perhaps in the story submitter's view, a substantial 77-90% increase in fps results in a marginal gaming advantage?
In fairness, it probably depends on the fps you're getting with a single video card; if you're trying to play at 15-20 fps I could see how doubling that might be an advantage. If you're going from 60-120 fps... well, I don't think that would help my gaming much.
Ok I used to be one of the lucky ones to have tried and seen 3Dfx voodoo2 SLI in it's peak days. I was beyond impressed, though the performance was no where close to 77 percent.
To publish a number like 77 percent, do people realize how insane that number is. I'd have to get rid of my frequently overheated ATI 9800radeon pro card now.
It seems to me the simplest solution would be to do as Alien does, and split the screen in half. In this way, all functions could be split. You'd have some effort duplicated at the border, but that's not a big deal as long as overall PCIe bandwidth isn't maxed out.
Under this config (and with proper drivers), each card is not only doing half the resolution - but could be working with a scene that has significantly fewer vertexes. Assuming this splitting can be made to work at all, this seems like the way to go.
Let's not stir that bag of worms...
Alienware took a very different tack with their solution because it requires a 3rd PCI slot AND it's analog (3rd & 4th pics). I guess its a series of tradeoffs: Space vs flexibility, with Nvidia winning the battle for space but losing on flexibility.
That aside, its rediculous that nvidia is expecting their OEM cooling solutions to do any kind of justice to the heat from those cards. Alienware already expects water cooling to be part of the solution and has cases designed accordingly... couldn't NVIDIA have done it any other way? Do they absolutely have to have a hardware link between their cards?
"A power draw of 250 Watts for the 6800 Ultra SLI solution is very realistic."
Then explain how this will work.
[Fuck Beta]
o0t!
So now they can catch up with ATI's latest Radeons?
If Yoda so strong in Force is, why words in right order he cannot put?
No, in '98 I had a great job and salary.
Sanity is the trademark of a weak mind. -- Mark Harrold
CRT > LCD / PLASMA
This equation just made me laugh. Like...if plasma is less than one, just think of how much smaller LCD has to be than CRT!
OK, I'm done.
They don't make the top-end solutions so people will actually buy them :-)
They make them so people will think they've got the fastest product out, and so therefore are the best brand, and so people will then go out and buy a moderately-priced card of the said brand.
ATI has gone ahead in leaps and bounds of marketshare because their Radeon 9700/9800Pro cards have been faster than nVidia's best. Demographically, hardly anyone buys these high-end cards, and only a couple percent of their profits are made out of these things, but people think whomever has the best out at the time is the company to buy from.
nVidia knows this release will make it to computer magazine reviews. Novices will read it, think nVidia is good, and go out and buy a 5200fx. This is how it works.
Who thinks this is a problem? If you don't want to buy two of these cards, here's a novel idea: don't. Let those who want to burn their money buy two of these cards and get the nice placebo ego-boost and the rest of us will exist quite happily. Why bother criticizing these cards or the people who will buy them? If you aren't one, why do you care?
http://xkcd.com/386/
Athlon64/FX/Opteron platform on the way. No need for pricey Xeons.
http://www.theinquirer.net/?article=16874
Isn't the reason that trees are shown as a couple of bitmaps due to hardware limitations?
Seriously, how the hell d'you expect anyone to power this thing?!
The 6800 requires at least a 400W PSU IIRC... does adding another GFX card mean you'll need another 300W for the extra card?
Moderation Total: -1 Troll, +3 Goat
Alienware already has a patent-pending process to do SLI on their own motherboards, whether it is with an ATi or Nvidia based videocard. The two caveats are: 1. so far, this will only be through Alienware, and 2. the videocards have to be exactly the same card.
e ss _release_template.aspx?FileName=press051204.asp
Alienware purchased a former 3dfx licensee who had outstanding patents on some of their own SLI tech. Alienware has wisely furthered the research and will be marketing it soon. And it doesn't require a Xeon processor...
Here's the press release:
http://www.alienware.com/press_release_pages/pr
"Right now, somewhere in this world, Scott Baio is plowing a woman he doesn't love," - Peter Griffin, *Family Guy*
In other news, the power supply folks at Antec, Enermax, and Zalman just announced an increase in expected earnings.
NVidia's next chipset for the the Athlon 64/Opteron/Sempron series of processors, the NForce 4, will allow for 2 PCIX-16 slots. Spiffy, huh? So it looks like they will be able to corner the market for the very top end stuff with this in combination with their new chipset. Unless Intel releases a desktop chipset with same capability. Or Via. Or SiS. Or...
For the simple reason being that nvidia's solution is the solution by the video card creator. Where Alienware's solution is a solution by a PC maker. One would think that the nvidia solution would be much faster and more cost efficient than the alienware solution.
Why doesn't nvidia and/or ATI just design a dual GPU video card with 1GB of memory running on a single 16x pci express slot? The bandwidth should be enough in that slot right?
Really, the biggest application for this kinda power that I can forsee would be game developers who want to see how well their games scale for next-gen video hardware...
Except that the next-gen hardware might be more than just a faster speed for pumping out polygons at greater resolution, but can include a newer/better shader or some other technology that makes the visual experience more realistic.
For 3D applications like Maya, Softimage, etc., a few gigs of memory on the graphics board would be more useful than this. For games, putting a TV-type resolution doubler on the back end, and rendering at half resolution when things are busy, would probably look as good. When you're running at some huge resolution, moving images would have a bit less resolution, but when the image stabilizes, you'd get full resolution.
There still are no nVidia drivers for Fedora Core 2.
They are talking like 2 more months!
Ahhh, what sweet memories.
I bought a shitload of 3DFX stock back in the late 90s because they were the king of 3D. I remember walking into a computer store, and seeing something on the screen... I thought it was clip from a movie, but they told me it was Mechwarrior 2 (I think 2) playing on a Voodoo card. My mind was blown. How they got movie-like graphics onto a computer was beyond my capacity to understand. I dropped the $350 and bought one immediately and played with it and loved it.
Then, after a while, I thought, 3DFX is the king and they will never die. I put my money where my mouth was and forked over my entire savings to buy 3DFX, around $15k. There-in I learned a few great lessons:
1) The best technology doesn't mean the best company. "Good enough" with a better run company will usually blow you away. Ask Microsoft or nVidia (well, at the time nVidia wasn't the top runner that it is today).
2) No matter how great of an explanation you make, the stupidest things like 16-bit color vs 32-bit color can kill you (22-bit color just doesn't cut it to the dumb-ass consumers). It's better to just cross your t's and dot your i's in the first place so that you don't have any such vulnerabilities.
They went tits up, and I basically lost my money. nVidia bought the remaining pieces of 3DFX, and that includes all their patents. I'm not surprised they went SLI, and for companies that use it like 3d effect companies, it will probably save them bundles of time.
And believe it or not I wish I knew a more pithy way of saying that. Shi, is that even grammatically correct? Anyways-
This entire thing is really an exercise in besting the competition. In this case, ATI and in this case, they are badly in need of being able to say that they are #1, because ATI stole the crown and hasn't really given it back just yet. This whole mess is really just an exercise in brand building/marketing. Just trying to get people to think of nVidia as the best/etc-- its aimed at those who can only afford one card, trying to influence their overall feeling for nVidia.
nVidia's 3DFX property didn't have much competition back when it first introduced the voodoo sli.. ahh how times have changed.
...unfortunately no one can be told what The Mat^H^H^HGoatse is...they must experience it for themselves...
Over a month ago Alienware unveiled its dual video card solution, however it doesn't connect like Nvidia's array does.
"Alienware is using an exclusive software solution as well as a video merger hub. Both solutions are patent pending and were developed in house by Alienware. In addition Alienware has developed a dual PCI Express graphics slot motherboard (X2). This motherboard is exclusive to Alienware and is also patent pending."
Alienware's Video Array splits the screen into two regions rendered separately by the cards, like Nvidia's concept does.
Also noteworthy, according to Alienware SLI doesn't function the way the Hardware Analysis article suggests - they say "SLI is a technology developed and used by 3dfx, where each card rendered alternate lines (one card would render the odd lines, the other card would render the even lines)." Nvidia's dual-card setup is not SLI in the sense of the old alternating scan lines.
Given that this configuration requires a Xeon based system w/ the dual PCI-E slots, this seems geared more towards the 3D development end of things, with Maya, Softimage and such. I've yet to meet a gamer-only with a Xeon rig, so this would seem to be a boon for the new Gelato systems, allowing for more GPU power. I just hope Nvidia doesn't end up emulating 3DFx's later moves in which it decides raw speed > innovation, as that is not really a winning strategy, especially these days where we're on the brink of a new age of gaming graphics using advanced shading techniques previously only seen in pre-rendered footage.
"What can a thoughtful man hope for mankind on Earth, given the experience of the past million years? Nothing." -Bokonon
I'm just gonna say NO to this.
I just threw away my 6800 assembler book!
-Mike
two pci-x slots are used, but four are taken up by the gargantuan cards. thanks nvidia.
lose != loose
If you're getting 120fps and this rig pushes it up to 240fps, the increase in your score and enjoyment of the game is far smaller than if the jump was from, say, 20fps to 40fps. No matter how hardcore you are, there's always an upper limit beyond which the additional frames simply don't matter, and this combination will push any game on the market well beyond it.
The benefit of going from 60-120fps would be that you could then knock the resolution up a notch or two, and get 45fps at 1600x1200. On the other hand, if you're already getting 60fps at 1600x1200, you could then turn up anti-aliasing, and make the game look good, as well as run fast.
I miss having three video cards in my computer, using up all my precious PCI slots.
The average PC motherboard has everything onboard. You won't need cards for firewire and USB, they're onboard. You might need to put in the audigy, but your system only need have five slots for that.
:)
And the average PC motherboard is a piece of shit, and not what you're going to be using with this setup.
I know onboard has improved a lot in the last few years, but I still want a LAN card that can do bus mastering, for one. Not to mention that even if your motherboard has all the extra USB/firewire ports built in, you still need slots to put the extra connectors. Front of the case notwithstanding.
Tt's hard to be sure since the faded backgrounds make me think all those pictures are PR. It seems like in the end, the high-powered cards that would really blow us away will take up two slots apiece.
This will suck for two reasons:
This won't matter as much if you get a slimmer card, but it still doesn't seem like there's much room for a waterblock, even if you wanted one.
[Fuck Beta]
o0t!
Gizmodo reports that Apple has announced new aluminim displays to match their G5 PowerMacs. Along with the 20-inch ($1299) and the 23-inch ($1999), a new 30-inch mega-display has been announced that will be driven by a customer dual-DVI video card that will cost $600. I'll get you pictures as soon as I can.
The 30-inch display will cost $3299 without the necessary $600 video card and will be available in August. Unconfirmed resolution is 2560 x 1600.
Looks like the video card is from Nvidia, which would explain that announcement earlier today about SLI, I bet.
Tom's Hardware brought up an excellent point: What happens to hardware pixel shaders that refer to a pixel that wasn't rendered by the card it's running on?
Given the almost limitless flexibility on shaders, this strikes me as a possibly huge performance bottleneck that may even make the SLI solution slower than just one card under certain conditions - or just render inaccurately.
Throx
Fear: When you see B8 00 4C CD 21 and know what it means
Nvidia goes for Nforce 4 PCI Express, all K8 support and Sound Storm 2 By Fuad Abazovic in Wien: Monday 28 June 2004, 12:07 NFORCE 4 will be the name of the project that will finally bring PCI Express to the Athlon K8 market. This chipset will be socket 754/939 ready and 940 compatible and all CPUs Athlon 64s, FX, Sempr0ns and Opterons will work on it. The platform will have one to two PCI express 16 slots making this Geforce 6800 SLI mode possible for not quite as much crazy money as you'll have to invest in Tumwater and Xeon CPU. The good news for Sound Storm lovers is that this chipset finally features these nice audio capabilities. It's now called Sound Storm 2 - no surprise there - but we don't know anything about its features as yet. Nvidia PCI Express stuff was scheduled for Q4 and it hasn't changed from that so far. We wrote Nvidias C8K-04 captured in living hues and now we know that this is the Nforce 4 project. Last time we asked Nvidia's Drew Henry director of this chipset business about Nforce 5 for Intel CPU's he said laughing where is Nforce 4. This is your answer Drew.
I would love to jam a monster like this into my shuttle. It's the fastest coolest little machine I've ever owned. Some problems dual booting into debian on it, but I think I've got those mostly ironed out at this point.
So, is it a matter of it not working AT ALL unless there's two PCI-x slots? And for that matter unless there are TWO of them? Of course that would leave the SFF crowd out.
But seeing as AGP is merely a modified PCI slot, I wonder if we'll see an AGP version of this card. And for that matter if this new card sort of works like the old (mindblowing at the time) voodoo2 cards where if you had one, it was great and if you had TWO it was brain-shattering. Or is it an absolute MUST that there be two?
And lastly, what about heat on these suckers? Doesn't look too good for an SFF machine, but I would love it if it did work. The portability to gaming sessions alone is of immeasurable value to my SFF enthusiasm. I would however, depending on how much expendible income I could free up build another big-box around this system. Not just for gaming, but I think these cards would give the old Oxygen cards a run for their money for Maya. And oh yeah, not that I'm holding my breath on this one, (I'm sure this has been asked elsewhere) but what about linux drivers?
Quod scripsi, scripsi.
I started putting in the flash blocker because the ads were so obnoxious and distracting that I had to put my hand over them to read the content.
I don't block static banners.
is it that the card is used to full capacity with plenty of room on the Bus.
Yes.
Call it the Mongol Horde approach (or, for Slashdotters, the Beowulf Cluster Graphics Array -- or maybe some more catchy acronym along the lines of RAID).
Take a load of cheap cards -- for instance, one of the many low-end cards available -- say, nine of them (nice square number: 3x3 array), and have each assigned to one tenth of your display, then have something downstream tile them back together, and bingo, super duper card. Better yet, design a single card using the cores of the nine low-end cards, which would no doubt be cheaper than nine times the cost of the single card (not to mention not requiring nine free slots on your motherboard(s) (!)), at which point maybe you'd bump it up to 16 or 25 cores.
Is there some fundamental reason this would not work (besides the basic issue of splitting and recombining your display)?
"A great democracy must be progressive or it will soon cease to be a great democracy." --Theodore Roosevelt
How much are they going to charge for the little SLI connecter doohickey? I see a $50 premium to "let" the consumer use this SLI mode.....
Using the fact of PCIe's long cable run lengths, I preicted the return of SLI over a year ago.
In fact, I further predict that graphics rendering will be externalized, and high-end rendering boxes consisting of just a card (or two) connected with external PCIe cables will be stacked and connected to a central host for extreme performance.
Why keep those hot, noisy cards inside the main CPU box?
Da Blog
1. Make your own corded solution 2. Sell it. 3. Profit!!
I have a hard time seeing how this works on modern hardware TnL cards. Vertex shaders, for example, can't be parcelled off to one card or another based on which half of the screen the vertex is, because you don't know where it will end up till the vertex shader is run (the shader can move the vertex to anywhere on the screen). I can see this working at the pixel shader level - give each card half the screen pixels to work on - and that's probably why it would have worked really well on Voodoo era cards, since all they did in hardware was rasterization, which could be split between as many cards as necessary. But modern cards do vertex transformation and lighting in hardware as well - I don't know how they can split vertex-level operations between different cards.
I guess this makes spending $5k+ on that new Alienware ALX 100% pointless now doesnt it! Thank you nVidia!
AnandTech has a more thorough review
2 dual core cpu's
2 hard drives in raid
2 gigs of ram
2 monitors
and now 2 graphics cards
arielb
I really don't have $1000+ to spend on just a video card... (Well, technically two)
Please flee in terror in an orderly manner.
You voted for Bush, didn't you. Ok... I'll type slow so you can read this...
I am NOT going to click on the banner add regardless of its content. So how does hiding it cut revenue, when that potential revenue from me, in a best case scenario, is zero?
push any game on the market well beyond it.
ITYM "push any game on the market NOW well beyond it."
Ewige Blumenkraft.
Or, Redundant Array of Very Expensive Graphics-Cards.
This is great news! I knew that there was work on implementing RAID for memory and CPU's, but this is outstanding!
Now my gaming need never be interrupted on the off-chance that my video-card blows during an intense Far Cry session.
I'm hoping to buy six of these cards so I can implement RAID-5 across five of them and have a hot standby.
Why isn't Creative looking at doing this for sound cards?
Seriously, why can't the money used developing this stuff be put into VR again? For FPSs, it provides *way* more realism than increasing pixel count and frame rates ever will.
With OLEDs supposedly about to revolutionise the display market, could we have another stab at that technology?
This is a clip from the apple web site
Dual Link DVI
The 30-inch Cinema HD Display requires the next level of DVI connectivity -- "dual link" to drive the massive amount of pixels to the screen. And the NVIDIA GeForce 6800 Ultra DDL graphics card (available from the Apple Store) delivers, with the most advanced graphics engine available. This card, designed specifically to support the dual link DVI connection, delivers 2560 by 1600 resolution. Even better, it can drive two 30-inch Apple Cinema HD Displays, giving you the ultimate creative canvas. This card will be available for Mac only in August 2004.
Not sure if Dual Link is the same thing as SLI, but with a different name.
So now it turns out that'll I'll need to sell BOTH my kidneys to be able to have the l33test video card setup in my street.. well for 3 or 4 months anyway... then i'll need to start harvesting organs from whinos in order to upgrade to the GF7900 Quad-SLI or whatever is the next big thing
External rendering boxes are nothing new
That's true, but now us common masses will be using this for the likes of FPS fests. I see people stacking their rendering boxes along with their eSATA drives in nice, consumer-friendly racks with Ikea styling.
Da Blog
You could do SLI with the TNT and TNT2 as well, with similar performance gains. It never caught on. And they were using standard PCI slots.
seriously?
i ent&ie =UTF-8&oe=UTF-8&q=define%3Aturbofan
"gas-turbine engine with a large diameter cowled fan"
http://www.google.com/search?sourceid=navcl
They're gonna need a 5 row aluminum radiator to cool it! Don't forget the high flow water pump. I'm expecting to see them balanced and blueprinted, knife edged, full roller, ported, polished, nitrous injected, and blown. Damn... I wanted to say bumpstick in there somewhere... hehe bumpstick...
It doesn't matter if you use the thing as a skateboard ramp, the fact is you bought a very high end computer. Unless you got an unrealistic deal on the FX53 (ie around $650), any reasonable business would have gone with a slightly lower end machine (say, a regular Athlon64 and a Radeon 9600) and upgrade every 2.5 years instead of every 5 to support medium-to-high-end applications. It just makes no sense to have an incredible pixel shader to do photography editing and AutoCAD - you get no value out of being ahead of the curve for this kind of application. A part half the price would give the same performance, and allow you to buy another much better part in two years.
But you didn't buy a mid-range computer, because you are a high-end computer buyer. There's no shame in that, and nothing wrong with it. It means you have a great computer right now - you just paid the gamer's premium for being a few months ahead of the curve.
And a current, maybe-slightly-higher-spending version of you is quite likely able to talk himself into buying a dual-video card monstrosity like this article talks about.
Let's not stir that bag of worms...
Well I can agree with that.
My new