Slashdot Mirror
Nvidia GeForceFX(NV30) Officially Launched
egarland writes "Tom's Hardware has a new article previewing the new GeForceFX chip and discussing its architecture. 0.13 Micron, 16 GB/s memory bandwidth, 128-bit DDR2 memory interface, 125 M transistors, support for 8x FSAA. Sounds like an interesting chip. They stuck with a 128 bit memory bus so ATI's R300 still has more memory bandwidth (19.8 GB/s) but NVidia has new lossless memory compression so we will have to wait for benchmarks to see if NVidia comes up a winner here. The reference card also sports a massive new cooling system which is worth a look."
Readers Oliver Wendell and JavaTenor add links to additional stories at The Register and at AnandTech.
That's nice. Now maybe NVidia will find the time to FIX THEIR FUCKING DRIVERS. Christ, they're becoming the new Diamond when it comes to shitty software.
Time to buy a Ti4600 :)
Actually, Anandtech has a benchmark that was provided by Nvidia showing a 40% increase over ATIs 9700.
NVidia's official Geforce FX site
NVNews has a large group of links to previews(scroll down to the "Geforce FX Preview" article)
Some impressive images from the release demos
I'm not so sure about that cooling system. Why put the intake right next to the output? Seems to me like it'll just be sucking that hot air right back in.
I'd think it would make more sense to use air inside the case and blow it out the back. With a grill/fan on the front of the PC, you're helping to improve the overall air-flow inside the system instead of just recycling your heat-wash.
AnandTech's coverage includes an nVidia-supplied benchmark that shows the NV30 beating the 4600 by 2.5x in Doom 3 (and the Radeon 9700 by about 40%). Of course, no one knows under what circumstances these benchmarks were obtained. I don't think any "independent" benchmarks will be available for awhile.
How many Watts does this monster dissipate?
:/
I'm just thinking of the power economics of the todays 3D accellerators...
So how many watts is this GPU drawing, to require an active cooling system that major? It seems that the latest GPU's from both major manufacturers are favoring a brute force approach to performance, rather than improving their architecture. I wonder what implications this will have for power supplies in your average PC - are we getting to the point that a fast P4 or Athlon system is going to require a 600 watt or more power supply to be adequately stable?
I also would love to hear how loud this video card is..blowers are generally pretty noisy.
There I was with my Beowulf cluster of GeForceFX(NV30) cards..
The duct tape glistened in the weak 40 watts of light in my parents' basement. "g1bb0r m3 T-Fl0p5!" I screamed but it was not to be. There was no joy in Mudville, the mighty cluster had blown a fuse.
Trolling is a art,
Firing Squad, also has some preview (with pictures) up here. From a technical stand point, seems to be pretty nifty, but I'm not convinced I'd rather have this card instead of 3 hours with a "massage therapist".
You don't need filters if the fan is always blowing out. Besides, there's always going to be dust in the box anyway, you just open it up and blow it out with Anti-static air. Looks like you could blow out that cooling unit too. That's a good thing to do every 6 mo. or so anyway, depending on the environment.
I could hook that thing up to my ductwork and save a fortune on natural gas this winter.
Trolling is a art,
Damn, Nvidia, why couldn't you have this thing ready for fall?
I've been searching for years for a leaf blower that could run Doom III at acceptable frame rates.
This board is clearly out of spec... since when I need to free up two slots to add a graphics card?
Obviously inserting it wont be easy and expect many breakage and damage returns.
NVIDIA has a few more shots of that Fairy:
1
2
3
Compression within graphics boards is very different than other kinds of compression. They aren't really trying to make the amount of data you need to store smaller; they are just interested in making the amount of data you need to shuffle between the chip and the card memory smaller. They also know that in some circumstances (multi-sampling) the data is going to be redundant in very predictable ways. This lets them take some shortcuts that let them have good average compression ratios, lossless, with very low latency. The risk of very bad cases is small--people aren't going to run games where everything looks like TV snow--and the worst-case penalty isn't too bad.
Another preview at HardOCP here.
I've read that nVidia has stopped GeForce4 Ti4600 production and is only selling the GeForce4 Ti4200 GPU.
In short, better get that Ti4600 card very soon, because they could be gone in a matter of months.
Dear Timothy,
1. Do you understand what the word 'launch' means?
2. Are you aware it is not yet February 2003?
3D graphics are fine and good, I do play enough games to want some polygon-smashing horsepower.
But has nVidia done anything towards improving 2D and multimedia performance yet?
The difference between the Radeons and the GF4's when it comes to watching DVD, using TV-Out, or just plain desktop computing is night-and-day.
The nVidia offerings always seem plagued with washed-out colors, shimmering refresh rates, albeit not nearly as bad as the 3DFX offerings. ATI cards have always been as good as it gets.
Sure I do alot of gaming, but not all of it is in 3D. I also watch movies, write code, surf the net, etc, etc.. Not only does nVidia never pay attention to any of that, nor do any of the review sites.
Video card != 3D Accelerator alone, IMO.
I don't need no instructions to know how to rock!!!!
There's this trend in computing to make everything faster, more featureful, hotter, and more energy consuming.
I agree. We're not getting huge, usable leaps in computing capabilities, we're getting continual, incremental improvements. Even these incremental improvements are not coming for free, we're getting them at the cost of increased power consumption, and millions of people throwing away motherboards and video cards every few years. And the incremental nature of it all keeps developers back a couple of generations. It's just barely getting to the point where you can realistically ignore everyone who doesn't have hardware T&L, several years after the introduction of the GeForce 2. But this is still a questionable choice, as a large number of PCs from Dell and Gateway still ship with generic video chipsets that don't have hardware support for T&L. Doom 3, which isn't even on the release radar yet (2003? 2004?), is the first game that's going to require the pixel shaders of the GeForce 3 and beyond. No other developer is going out on such a limb, as cool as shaders may be.
I'd love to see a quantum leap in desktop PC capability that isn't a one-to-one trade of MIPS for wattage. It's very possible, but we're running down this bizarre path where everyone gets all excited about a 9% increase in raw clockspeed (which translates into maybe 4% in benchmarks), even though it increases power consumption by 9% or more.
I'm at the point where I'd be willing to chuck the historic trappings of desktop PCs--x86, UNIX-like operating systems, C++, gcc, etc--for something simpler and cooler running, whose blatant wrongness doesn't eat away at your soul every time you use it. The whole Windows vs. Linux nonsense is a complete red herring in that regard.
I predict that we'll soon be buying big metal graphics controller boxes from nVidia complete with heavy duty power supplies and massive cooling capacity. After you get it home, you'll open up your graphics adapter and insert a little motherboard and CPU into an option slot to complete your computer system.
New hardware mentioned on Slashdot. Now it's time for all the lamers to come up with the following posts:
:) Maybe along with a dual Opteron machine. And before you scream excess, have you checked Pricewatch lateley? I remember paying $3300 for a single processor PII-300 with 64MB of RAM and a Riva 128 in January of 1998. If the Opterons don't cost that much more than the high-end Athlons today, I could put together this machine for significantly less than that!
1) Who needs all that power anyway? I'm running Windows XP just fine here on my 486SX/33!
2) Why cares if it's fast? It uses up too much power and has a *fan* on it. God forbid a computer have a fan on it! It sucks because it's not fan-less like my Mac!
3) Sure it might be fast, but I bet it isn't as *efficient* as a G4!
4) NVIDIA sucks because it's drivers are closed source.
Did I forget anything? Anyway, I couldn't care less what the lamers think. This is a genuinely cool piece of hardware. There are a few things that make it so:
1) 500 MHz! That's half a gigahertz! A very large jump in clock-speed here, much more so than the usual 33 MHz pussy-footing the industry (particularly Intel!) is guilty of.
2) Compressed-memory access. Ah, computational power exceeds memory bandwidth to the point that it's more efficient just to compress the data before sending it over the bus... The 16 GB/sec memory bandwidth (which is also quite a big jump from existing machines) is made even more impressive by a lossless compression that can achieve 4:1 ratios. This is very helpful for multisample AA graphics, because it reduces the memory bandwidth hit to just the pixels that occupy the edges of polygons rather than every pixel in the scene.
3) Fully floating point pixel pipelines. Carmack was asking for 64-bit floating-point point pipelines a while ago. While this doesn't quite get there (it's 32-bit floating point) it is a major step, and makes life a lot easier for game developers.
Overall, this card is definately in the cards for me
A deep unwavering belief is a sure sign you're missing something...
NVidia has a list of "Lauch Games" for the GeforceFX. Command & Conquer: Generals, Unreal II, Rallisport Challenge, Sea Dogs II & Splinter Cell. Screen shots and some movies are included.
http://www.kubuntu.org/
Dust doesn't hurt chips, but it does insulate them which can lead to excessive heat which does damage chips. Filters on cooling fans is a bad idea, simply because having a filter will increase resistance and reduce airflow which kills the desired cooling effect.
:) ) It's nice to see that nvidia is thinking of these things.
Instead of using a filter simply buy either:
1: A can of compressed air every now and then (expensive, but easy and reliable)
or
2: A small air compressor (however this can get much more expensive in the short term especially considering you need not only a compressor, but also, hose, fittings, an air chuck and most importantly a dryer (aka de-humidifer), so unless you have alot of stuff that needs cleaning and you live in a place that makes it needed fairly often you should probably stick with #1)
I must say though, what a cooling system! I don't know about everybody else, but I used to have a nice voodoo 3500 that would get so hot that you could burn yourself on it, I was always worried about that thing.... I finally rigged up a cooling system for it (yeah I know, buy one.... but it's more fun to make it out of old parts
I am the only one with my AGP slot as the first slot at the top on my motherboard? That means there's no open slot on the back of my case for that fan to stick out of. The only way I can see a contraption like that working is if it was taking up two PCI slots, which of course it doesn't... Any ideas?
Well, I've had my GeForce2 for almost 2 years now, and with this announcement of the GeForce FX, it's finally a sign to upgrade.
:)
It's funny, practically my entire workstation (P4 2.2GHz, 256MB DDR400, 80GB HD, etc.)has been upgraded in terms of components, however, my video card has remained static. Not that I'm complaining, because I can run pretty much every game out there at (what I consider to be) fairly decent speeds. Take Age of Mythology as an example. It's more than fast enough. Unreal Tournament 2003 is a tad different, as I have to turn down some of the graphics, but it's is still fine for the 'average' game. Plus, my Xbox and PS2 are for my gaming needs
Now, does the theory of diminishing marginal utility apply to video cards, or is it the opposite? How much more powerful can video cards get so that we won't even 'notice' (at least in the loose sense) any difference when playing games? The Radeon 9700 Pro (with a fast CPU) can run pratically every game on the market at max details at most resolutions. Well, so can the GeForce FX 5800. Sure it may be 30-50% faster, but the utility gained for current games is definately marginal.
Since I've held out for 2 generations of video cards, for me, it's definately the time to upgrade. Though, it's not really because my video card is too 'slow'. I suppose it's an issue of just gloating to my friends!
Moreover, in terms of approaching cinematic rendering, nVidia is definately going in the right step. They are quickly approaching the level of "Final Fantasy" in terms of quality of output. Nonetheless, they'll still need to add quite a bit of horsepower to be able to do it all in real-time.
So there you have it; the elusive NV30 has surfaced in the form of GeForce FX. ATI has won the first round with the Radeon 9700 Pro, what will be most interesting will be what ATI has up their sleeves when the GeForce FX hits the shelves in February.
Myself, I had a GF3 Ti500, I upgraded to a GF4 4600, but it wasnt much faster, returned it. Then a couple games came out (Battlefield 1942, Unreal2003) that really needed some gfx horsepower. So I bought the Ati 9700, Amazing. I can run older games with 6x AA perfectly, and Newer games run at 60FPS with 2x AA enabled. The GFX card works fine with the CVS version of Xfree also. (Or vesa mode for older 4.2.1) Also, I can output to TV at 1024x768, and have it mirror my monitor, great when playing some multiplayer games, or playing some divx/svcds. The Ati 9700 is a very nice product, and found some great forums at Rage3d for questions and updated beta drivers. (Like the new DX 9.0 drivers and DX 9.0 demos)
My goodness, can you imagine a "workstation" running one of these nVidia cards with dual Itanic processors? Heck, if you got a university to run this configuration, you could bring Enron back from the brink. I see 20amp fuses in many homes going "POP" right now.
In a desktop the reference design is already FUBAR - the output vent is below the intake vent.
Oops.
You can't suck in air from the case because you can't be sure that there's enough ventilation to let you suck the air in -- you always want to maintain an equal ratio of input and output airflow. The only way Nvidia could do this is to put the intake and the output on the card itself, which leads to the situation we see currently.
Preventing the output being sucked back into the intake is pretty trivial though - take a piece of cardboard and put it between the two. That will solve the majority of the problem. Yes, it's inelegant. But if the cooling problem has gotten to the point where you need a heat pipe with a blower separate from the rest of the system then you're pretty much SOL on elegant solutions anyway.
That new graphics card sure looks pretty and EXPENSIVE with all that copper. This will certainly add to the cost of that product. I wonder what percent by weight of the entire product is copper, seeing that copper is a commodity metal.
Regarding those comments about the cooling system not having a filter, this is a pre-production model. Give it some time, it will have to use a filter to keep the small space between the copper fins free of dust.
Hey Bob, while you're out at Murray's Automotive, get me a new oil filter model number P3160 for a Saturn SL2 dual overhead cam and FX160 filter for my NVidia graphics card, 128MB DDR2 RAM, and be sure to read the serial number information. My FX card is post 4375XXX, so it doesn't need a finotany rod or a muffler bearing.
From a developer perspective, we're headed for a shader fight between NVidia's Cg, OpenGL 2.0 shader languages (shader assembler, ISL, and Quartz Extreme) and Microsoft's HLSL. It's not enough to have shader languages; they have to be supported in the content creation tools, so the artists can see what they're doing. This will take a while.
Developers need to buy this thing, but everybody else can wait a year.
years after the introduction of the GeForce 2
2.5 years to be precise. The GF2 was released in May, 2000. I wound up having to buy one the 2nd day it was out, so I remember (old V2 setup wouldn't work in new system).
Doom 3, which isn't even on the release radar yet (2003? 2004?), is the first game that's going to require the pixel shaders of the GeForce 3 and beyond
Doom3 is allegedly scheduled for Christmas of 2003. I'd be surprised if they missed that, but id software is usually more focused on getting it done right than on time, so who knows.
As for the features - by that time everyone will be going out on the same limb. As usual, the D3 engine will be licensed by many people and all those games will require the same level of hardware. D3 will take advantage of most of the features present in the GF4/GFFx as well, so now we're back to the games being only a year behind the hardware.
I'd love to see a quantum leap in desktop PC capability that isn't a one-to-one trade of MIPS for wattage
Well, I have no idea what the power consumption of the GF Fx is, but it's not a 1:1 trade of speed to MHz - the GF Fx runs at a 500 MHz core, which is roughly a 40% improvement over the Ti4600. For that speed improvement you get (allegedly) up to 400% of the speed. Not bad.
Realistically, though, you've got to be kidding. Science and technology rarely deal with sudden massive jumps in capability or performance. It's all building blocks. If you want a sudden massive jump then you have to skip a few iterations.
Did I mention that I'm still using the aforementioned GF2? Yes, I'm looking to upgrade right now and I do expect a considerable leap in capability and performance.
I'm at the point where I'd be willing to chuck the historic trappings of desktop PCs
So vote with your wallet and stop buying stuff you don't need. The only blatent wrongness is in buying crap you don't need and then whining about it being evil.
It's similar to how you can't feel the air blowing towards a fan intake as well as you can feel the air blowing out. Try it with a household fan sometime. Orient your hand parallel to the intake/output so that you're not blocking the flow much.
So, if they can get the cool air from outside, it's a better solution than using the pre-heated air from in the case.
Can't you see that everyone is buying station wagons?
Its amazing!
The specs for this board should include a noise dampener to counter the hoover that they have strapped to its circuit board.
The ex-3DFX engineers that NVidia acquired somehow managed to brainwash the NVidia guys into releasing a gigantic monster of a board that can only rival the VooDoo 5000 in its unpracticality and ungainliness.
Those 3DFX guys have had their revenge.
Right now I'm on a project where we are reluctantly (well, I'm reluctant: others are quite happy) using SGIs: we just dropped mid-five figures, and will probably come close to six before we're done (on this machine, we have about another $500k or so worth already). A lot of this is because of SGI's graphics pipe: we're doing some convolution and other stuff where we use pretty much all of the 512MB of texture memory that we have.
I believe that current Nvidia Ti4600s have 128MB (256?) of memory, so I hope that a professional level of this new card might scale to the half Gig we need.
Additionally, the SGI is 12-bits per color channel, which is a bummer since the interface it is simulating is 16-bit monochrome. Sure, you can try and do tricks, but from a quick glance over the FX's specs, I see 32 bits per channel, which would be very nice.
With this FX card, a reasonably setup AMD Clawhammer system, and the scalability and preemption stuff that's going into 2.6/3.0 Linux kernels, we might be able to move from SGI within the next year or two, thus saving taxpayers on the order of $40-80k or more per system. A lot of development is already done on Linux, but it sure would be nice to move over fully.
Why didn't they name it the GeForce5? That sounds soo much cooler than FX. FX doesn't sound powerful at all, especially when their low end chip is called the "MX." Pronouncing the two isn't that different, too. Which sounds faster: Radeon 9700 Pro or GeForce FX? Sheesh.
"NVIDIA has hinted at offering another version of the GeForce FX at a lower clock speed that would only occupy a single slot cutout, but we will have to wait until the product line is announced before we can find out what the differences will be. Our initial guess would indicate that a simple reduction in clock speed would be enough to go with a more conventional cooling setup."
And:
"The other issue that users may have is noise, luckily NVIDIA has taken steps to make sure that the GeForce FX is one of the most quiet running cards they've ever produced. Borrowing technology from their mobile parts and combining it with the FX Flow cooling system, NVIDIA is able to dynamically reduce the speed of the fan based on the graphical needs of the system. When sitting in a 2D situation the card will scale back the clock speed of parts of the 3D pipeline that aren't in use, thus allowing the fan to spin much slower. As soon as you start using the GPU for games or any other 3D intensive applications, the clock speeds up as does the fan. The idea is that if you're gaming you're not as concerned with noise as when you are typing in Word."
Link: http://anandtech.com/video/showdoc.html?i=1749&p=6
"Your mother sent me here to kill you..."
- "Bill Cosby - Himself"
I don't quite follow. Wouldn't even-numbered ones be the new cards to buy? Give everyone else a chance to mess with the bleeding edge when they come out with a GeForce 5 or 7, and then snap up a technology-perfected GeForce 6 or 8 for half the price of the original 5 (or 7) series, with more features, more memory and many small improvements.
I've done this for the GeForce series up until now, sticking with my Riva TNT until GeForce 2 came out and then keeping my GF2 until I could afford a Radeon 9000 (which is a GF4 equivalent). I've always been happy with my affordable, yet cheap graphics performance (my last three cards have been less than $100 apiece).
Why didn't they name it the GeForce5?
...errr...hey wait a minute...
Remember that company called 3DFX whose "last" card was the Voodoo5? Then a powerhouse called Nvidia took over as highend "King of the Hill".
Funny how that works, eh?
I mean it is not like Nvidia has anything to worry about with ATI taking the performance cro
.
Have you read the moderator guidelines? Well, have you, PUNK? (and I want a Karma: Gnarly option)
I'm suprised noone has commented on NVidia's change to flip-chip technology yet. It's the first time that I've seen it used in consumer computer technology. Instead of having small legs like surface mount chips, the chip has blobs of solder underneath it, and the solder bonds to the PCB when the chip is pressed against the board during manufacturing. It's important because it lowers the capacitance of the external pins, which means that the chip can interface to the outside world at higher clock rates. It's an important shift in packaging technology.
Official MacOSX 10.2.7 Patch schedule
Because many new GPUs are reaching a stage where they are faster than our G4s, code has been added to swap the GPU into a CPU and the CPU(G4) into a GPU. We anticipate a 15-30% boost in Photoshop.