ATI's 1GB Video Card

Signify writes "ATI recently released pics and info about it's upcoming FireGL V7350 graphics card. The card features 1GB of GDDR3 Memory and a workstation graphics accelerator. From the article: 'The high clock rates of these new graphics cards, combined with full 128-bit precision and extremely high levels of parallel processing, result in floating point processing power that exceeds a 3GHz Pentium processor by a staggering seven times, claims the company.'"

Comments

use as a cpu?

[Toba82]

Why doesn't ATi (or nVidia for that matter) make CPUs?

They obviously could make some very powerful chips.

Re:use as a cpu?

[Tyler Eaves]

Because for most general computing tasks, floating point doesn't matter.

Re:use as a cpu?

[Kenshin]

I'm thinking:

a) Tough market to crack. AMD's been around for years, and they're still trying to gain significant ground on Intel. (As in mindshare.) May as well spend the effort battling each other to remain at the top of their field, rather than risk losing focus and faltering.

b) These chips are specialised for graphics processing. Just because you can make a kick-ass sports car, doesn't mean you can make a decent minivan.

Re:use as a cpu?

[TheRaven64]

Building a GPU is trivially easy relative to building a CPU. Here are a few reasons why:

• You have an OpenGL driver for the GPU and a JIT for shader language programs. This means you can completely throw out the instruction set between minor revisions if you want to. An x86 CPU must mimic bugs in the 486 to be compatible with software that relied on them.
• You have an easy problem. Graphics processing is embarrassingly parallel. You can pretty much render every pixel in your scene independently[1]. This means that you can almost double the performance simply by doubling the number of execution units. To see how well this works for general purpose code, see Itanium.
• The code you are running is fairly deterministic and unbranching. Up until a year or two ago, GPUs didn't even support branch instructions. If you needed a branch, you executed both code paths and threw the result you didn't need away. Now, branches exist, but they are very expensive. This doesn't matter, since they are only used every few thousand cycles. In contrast general purpose code has (on average) one branch every 7 cycles.

GPUs and CPUs are very different animals. If all you want is floating point performance, then you can get a large FPGA and program it as an enormous array of FPUs. This will give you many times the maximum theoretical floating point throughput of a 3GHz P4, but will be almost completely useful for over 99% of tasks.

[1] True of ray tracing. Almost true of current graphics techniques.

Re:use as a cpu?

[dcapel]

You, like almost everyone else seeing stuff like that thinks that the first time.

The reason why GPUs aren't uber compared to CPUs is that CPUs are general purpose, while GPUs focus on a few key areas. This allows them to get really good performance at a few things, but suck at most other things. Its the difference between a Jack-of-all-trades and a specialist.

Re:use as a cpu?

[moosesocks]

I understand that CPUs and GPUs have dramatically different roles.

However, how difficult would it be to write an operating system that offloaded floating point operations to the GPU, and everything else to the CPU.

Seems like that would be making the most efficent use of available resources..... (then again, isn't that the idea of the Cell processor?)

Re:use as a cpu?

[lostchicken]

You're absolutely right on, but I've got a minor picky point about the first part. While the instruction set must remain stable pretty much downrange forever in a given architecture, this is only really true for the front-end instruction set, which isn't what the CPU actually executes on any current IA-32[e] design. Only the decode stage of things is bound by the instruction set, and the actual bulk of the work is done in the internal architecture which can, and does, change between different designs.

Re:use as a cpu?

[VoiceOfRaisin]

how is this insightful? the parent did not say their gpus are cpus, he asked why they dont make cpus. obviously they wouldnt be exactly the same as the chips they make now. duh?

Re:use as a cpu?

[Dorothy 86]

However, how difficult would it be to write an operating system that offloaded floating point operations to the GPU, and everything else to the CPU.

Because then it wouldn't be Windows, which is what the (majority) of the target market uses.

Re:use as a cpu?

[Segfault666]

... Just because you can make a kick-ass sports car, doesn't mean you can make a decent minivan.

http://personal.inet.fi/surf/porschenet/text/fut _plan/varrera/index.html

enjoy...

Re:use as a cpu?

[Anonymous Coward]

The overhead of transferring all of this stuff to the graphics card for processing and back would be way too much for almost all FP operations. Even on a general purpose CPU the overhead of fetching the data is usually more than the time it takes to process. Thats why there are caches. They exist to minimize this overhead.

However if you have large amounts of data and want to process it over and over, you can afford to transfer it to a dedicated processor. This also happens to be what this card is designed to do.

Re:use as a cpu?

[Andrzej Sawicki]

Well, ATI offers software that allows you to use the GPU for computing physics. The method for offloading CPU load is indirect, but there you have it.

Re:use as a cpu?

[pherthyl]

Building a GPU is trivially easy relative to building a CPU

Easier? In some respects. Trivially easy? Not quite.

In contrast general purpose code has (on average) one branch every 7 cycles.

One branch every 5-7 instructions, not cycles.

Other than that, good comment, I didn't know about the (lack of) branch support in GPUs.

Re:use as a cpu?

[pchan-]

However, how difficult would it be to write an operating system that offloaded floating point operations to the GPU, and everything else to the CPU.

Funny you should mention that. The Intel 386 (and up) architecture has built in support for a floating point coprocessor, so it can offload floating point operations. In the early days, you could buy a 387 math coprocessor to accelerate floating point performance. Then Intel integrated the 387 coprocessor onto the 486 series cpus, and today we just know it as "the floating point unit" (although it's been much revised, parallelized, and super-scaled).

As for offloading to a GPU, well... that's what we do today. It's called Direct3D, or Mesa, or Glide, or your favorite 3D acceleration library. The problem with this approach is that it requires very specialized code. It's not something that can be automatically done for just any code, as the overhead of loading the GPU, setting up the data, and retrieving the results would far exceed the performance gains. In only extereme cases does it pay off: the workload has to be extremely parallelizable, with almost no branching and predictable calculations. Basically what it ends up is that the algorithm has to be extensively tailered to the GPU. Even IBM has had major issues offloading general purpose operations to their special processing units, and those are much more closely coupled to the CPU.

Re:use as a cpu?

[Anonymous Coward]

However, how difficult would it be to write an operating system that offloaded floating point operations to the GPU, and everything else to the CPU.

The thing is, the main thing floating point is used for is 3D graphics -- the same thing these cards are already being used for. So in effect, this has already happened.

Re:use as a cpu?

[mycall]

You think 7x is fast, you should check out what ClearSpeed CSX600 is going to do 10x with AMD's coherence HT.

Re:use as a cpu?

[mycall]

if FPGAs impress you, VPGAs, PLDs and ALUs blow them out of the water.

Re:use as a cpu?

[empaler]

I think that's the finnish version of 'Bandwidth Exceeded'...

Re:use as a cpu?

[fbjon]

No, it's a 404: real linky

Re:use as a cpu?

[ajs318]

There doesn't seem to be a market for a new CPU architecture.

One could make a blisteringly fast, ultra-RISC processor that completes every R-to-M instruction in one clock cycle {read the instruction from memory on the rising edge ["tick"], let the logic matrix outputs stabilise during the high period and write the result to memory on the falling edge ["tock"]} and every M-to-R or M-to-M in two {an extra tick is needed to read the operand from memory and the intervening tock is wasted; unless you can read on both edges, then M-to-R can be done in one cycle, with the logic matrix stabilising during the low period. M-to-M still takes two and might be better "faked" using two instructions; M-to-R followed by R-to-M, which would keep the instruction set simpler}. You could even do several R-to-R instructions in parallel, if you had more than one logic matrix {or a partitionable logic matrix} and a register bank designed so as to support multiple independent reads and writes. But such a machine wouldn't be at all architecture-compatible with anything else, and you would need effectively to run a kind of interpreter or emulator all the time to deal with the instruction sets of existing processors.

And that is what the modern Athlon-type processors do anyway: a RISC processor is emulating an old CISC processor. I personally have a sort of gut feeling that this is unnecessarily slow: code that was optimised for the "outward" CISC instruction set may not be truly optimal on the underlying RISC processor, since it might occasionally have to take a diversion to do something which the CISC instruction specification required, but which was not necessary in this particular instance {e.g. setting a flag which will never be read}. Saving even apparently just a few clock cycles in this way might well result in a performance gain, especially if it can be done in a heavily-used loop -- particularly somewhere like the kernel, libc or the shell.

However, every new processor architecture -- or, for that matter, every new way of using an existing processor architecture -- requires recompilation of existing software. In the Open Source world, that's somewhere between trivial and a non-issue. But in the Closed Source world, it's a showstopper.

Re:use as a cpu?

[BurntNickel]

Funny you should mention that. The Intel 386 (and up) architecture has built in support for a floating point coprocessor, so it can offload floating point operations.

Actually, the FPU as an adjunct goes back on the Intel x86 architecture to at least the 8087

Re:use as a cpu?

[clydemaxwell]

because OP's 'obviously' statement seemed to imply he thought the current GPUs were a good marker of CPU quality -- i.e. he was impressed by the GPU specs, and so wondered why they don't make CPU's (7 times better than current) so GP was right in pointing out that it's not a very good indicator of CPU strength, and thus we have little reason to believe nvidia would make a good CPU. tada!

Re:use as a cpu?

[somersault]

the problem is that the graphics accelerators work in a very limited domain, so the gfx cards engineers can concentrate on the things that it needs to be fast at (floating point calculations, tranferring memory contents efficiently). Normal CPUs have a much wider scope, and while I'm sure the engineers that design/upgrade x86 processors do try their best to make the chips fast, they have to spread their work around, and make sure that every area is decent, rather than one area spectacular. Also graphics cards are fairly self contained, while processors have their motherboard/chipset (which for intel would include the memory controller) to dictate which type of RAM they can use, and how much bandwidth the whole system has etc.

The thing is that you couuuuld make an x86 that runs using GDDR3 etc, but it would be rather expensive, and nobody (well, no majority market anyway) is going to pay to produce that, if only a few thousand people can actually afford it. In time the costs will come down, but until then we common folk just have to stick with whatever AMD/Intel/Whoever are producing.

But anyway, the main point I made, maybe not in a very technically accurate way, was that it's easier to build something that performs well in one area, than to build something that does everything amazingly well (without costing the earth to buy it).

Re:use as a cpu?

[Zaatxe]

The grandparent is really interesting and the parent is really insightful. I agree with both of them. Therefore, why should ATI compare the processing power of this GPU with a Pentium if they aren't made for the same things?

Re:use as a cpu?

[Anonymous Coward]

As for offloading to a GPU, well... that's what we do today. It's called Direct3D, or Mesa, or Glide,

Mesa is a software only openGL implementation. It does the exact opposite of offloading to a GPU. You can replace Mesa in that sentence with "openGL driver".
And as much as I liked glide, it died what 7 years ago?

Re:use as a cpu?

[Anonymous Coward]

Just for clarification: in the beginning, 486 line still had the option for a coprocessor. The 486SX used a 487 coprocessor, while the 486DX had floating point enabled on the chip. Afterwards, the 486DX2 and DX4 came out, and the SX disappeared.

Re:use as a cpu?

[lubricated]

what the hell is a porche cayenne, if not a minivan with regular doors.

Re:use as a cpu?

[CastrTroy]

Are there any open source projects out there that utilize the specific features of the GPU. I imagine it would be great for things like video transcoding. Maybe it could even be used to compile programs. At the times when i'm only browsing the web or typing up a word document it would be nice to have the GPU working on something useful, rather than using my CPU cycles.

Re:use as a cpu?

[Cat_Byte]

That may very well be the most aweful, ugly minivan I've ever seen.

Re:use as a cpu?

[LWATCDR]

Short answer
Yes that is what the cell does. And it is already being done for the things that they are good at.

You have to remember that GPUs floating point precision tends to be limited. I am not sure that they are even IEEE single precision much less double. They are great for things like... Graphics and video playback. In fact they are used for that a lot. They are not all that useful for science and engineering tasks. One of the things that I hear every now and then that scientific computer users complain about how modern CPUS like the Athlon and PIV lack good double precision floating point performance. That is why Sparc, Power, Alphas, Itaniums are still around and used. They actually have better double precision performance than the X86 line does.
That being said. I have heard that Nvidia is pushing the idea of using GPUs for some physics modeling in games and I know that some science types are trying to use GPUs for the tasks they may be good enough for.

Re:use as a cpu?

[Trashman]

Can a stock Minivan do 0-60 in 5.4 seconds? Conversly, can a Sports car get more than 20 mpg on a highway? They are definietly not the same.

Re:use as a cpu?

[Anonymous Coward]

There was also the clock doubled SX2 (same as DX2 with FPU disabled)

Re:use as a cpu?

[steveo777]

Getting off topic, but check out Honda's Hybrid Accord. I salivate over it every time I think of it. 253HP Combustion engine and a low-RPM electic. It gets about 24city, 35hwy for mileage. Quite speedy and relatively efficient.

Re:use as a cpu?

[lubricated]

Can a stock Minivan do 0-60 in 5.4 seconds?
Porche Cayenne, and probably some of the other luxury ones.

Conversly, can a Sports car get more than 20 mpg on a highway?
Chevrolet Corvette, Honda S2000, Most of them can, except for the high end stuff out of Ferrari and the like.

Re:use as a cpu?

[Trashman]

Chevrolet Corvette, Honda S2000, Most of them can, except for the high end stuff out of Ferrari and the like.

I should've added "and fit 7 passengers."

Re:use as a cpu?

[default luser]

Because the Pentium 4 was never made for powerful floating-point thoroughput using plain old x87 FP instructions. To give it better floating-point performance, a 128-bit vector unit (SIMD) was strapped on. With it, the Pentium 4 can perform 4 32-bit floating point operations every 2 cycles.

This is the one part of the Pentium 4 chip that is very much like the parallel execution units of a video card. If you want to maximize performance of the SIMD unit, you want a simple program with packed SIMD instructions. The key point is, the video card can do that better.

Re:use as a cpu?

[Kenshin]

what the hell is a porche cayenne, if not a minivan with regular doors.

About $50,000.

Re:use as a cpu?

[Illbay]

AMD/Intel/Whoever

Uh, Cyrix?

Re:use as a cpu?

[NetRAVEN5000]

You're right, they could make some pretty powerful chips. But there's a few reasons why they don't. First of all, all their chips use vector processing, which is what makes them so fast. Vector processors are quite a bit faster than regular microprocessors, but they also cost more to make. Also, because it's a vector processor, it might not be compatible with other more common architectures, such as x86. Not to mention, you generally just don't need that much raw power. If vector CPUs were to become a commodity, that'd be great! But I don't see that happening since most people just don't need that kind of speed. Although maybe some gamers would. . .

Re:use as a cpu?

[triso]

That may very well be the most aweful, ugly minivan I've ever seen.

No, that is still held by the Pontiac Aztek, http://tinyurl.com/q2bch however, that thing looks like a UTE more than a mini-van to me.

Re:use as a cpu?

[2short]

"a) Tough market to crack."
    If you could make a CPU that was 7 times faster than everyone elses, I think you'd find the market incredibly easy to crack.

"b) These chips are specialised for graphics processing."
Bingo; Graphics processing is specialized, and in particular, *highly paralelizable*. Rather than your minivan/sports car analogy, I'd go for this one: In a given amount of time, a Greyhound bus will get 20 times as many people across town as a Ferrari. Asking why GPU makers aren't threatening the CPU leaders is like asking why Greyhound isn't doing well on the Formula 1 circuit.

Re:use as a cpu?

[Anonymous Coward]

Apple had a "AV" series, the 660 AV and the 840 AV. Those systems had a DSP on the motherboard. Adobe accelerated some PhotoShop functions with that DSP. Some other software publishers did the same with their products. Once the PowerPC Macintoshes came out, Apple said it was faster to just use the CPU ( meaning they couldn't afford to put a DSP on the motherboard of each system ).

Re:use as a cpu?

[Anonymous Coward]

Well actually I kinda like the look of the Aztek - almost a mad max / blade runner style one (I think they had one in Dark Angel in dark paint).

Personally the ugliest one I have seen is the Fiat Mutipla which is apparently very well designed for a family but ugly as hell. Witness it here then mod me down for making you look at such a thing!

Re:use as a cpu?

[Hamsterdan]

They are different animals...

A CPU has to rely on old 386/486 code (20+ years old technology) or some software won't make it. (Some people *need* their old software)

What I would do, drop everything before the P6 core (Ppro/P2/PIII) from the hardware (CPU) itself, and run those bugs/instructions using an emulation layer. Just ship a CD with the CPU, and install it on the OS so it can emulate those instructions.

I mean, If your software *requires* 486 instructions,and used to run on a 66 or 100Mhz 486 it can be emulated quite easily with modern processors. That's what I would do. (VirtualPC and Vmware emulate a *whole PC* in software and manage to run at decent speeds, most of the times much faster than a 486)

Less transistors, so either a cheaper unit, or a faster/more efficient one. (not to mention it won't need liquid nitrogen :)

A GPU on the other hand relies on its driver and the OS's API (DirectX / OpenGL). So they can throw out inneficient(sic) code from the hardware itself and leave room for new hardware.

Besides, modern GPUs are massively parallel Floating Point Units, they are not a general purpose processor such as CPUs.

My X800XT has 16 Pipelines, but it's already outclassed, cards are coming out with *way* more than that :)

ATI & Nvidia have enough to worry about each other, they don't need big old Intel & AMD as their competitors.

(What I would like to see would be an up-to-date AMD chipset. It's their CPU, so why don't they build chipsets?)

Seven times you say?

[Anonymous Coward]

...result in floating point processing power that exceeds a 3GHz Pentium processor by a staggering seven times, claims the company. P4's double ALUs result in integer processing power that exceeds a FireGL V7350 graphics card by a staggering seven times, claims the company.

hackers quote

[Kwiik]

"it has a p6 chip. Tripple the speed of the pentium"

Re:hackers quote

[Anonymous Coward]

RISC chip.

Re:hackers quote

[j79]

RISC is good

Re:hackers quote

[Anonymous Coward]

It has a killer refresh rate.

Re:hackers quote

[OBCENEIKON]

"Yo, I bet it looks crisp in the dark."

Re:hackers quote

[default luser]

Yeah. It's not just the chip, it has a PCI bus. But you knew that.

Not bad...

[spartacus_prime]

What market is it directed towards? I hardly imagine that typical gamers or animators would have the funds for such a card.

Re:Not bad...

[ksheff]

high detail flight simulators or virtual reality apps.

I would have loved having a graphics subsystem with a 1GB for texture maps when I was doing gl work 12 years ago.

Re:Not bad...

[KajiCo]

3D Artists, Game Developers, Scientists who do three dimensional simulations, Weather forcasting models, Engineers who do real time simulations.

They're not really for gaming as much as they are for developing stuff.

Re:Not bad...

[AgNO3]

Well is is a FireGL card. and Hell yes animators would love this card. The only problem with this card is that it is not an NVIDIA card so right animators won't want this card.

Re:Not bad...

[AcidLacedPenguiN]

its less about market direction and more about making their insanely expensive cards from 2 months ago less insanely expensive.

Re:Not bad...

[RingDev]

This is a workstation card, not a gamer's card. 1 gig of memory on a graphics card will not help any video game on the market (even the 512 cards are over kill). Like the FireGL cards, they are not all that great for gaming, but are extremely impressive for rendering. Great for 3-d content creation, scientific modeling and other rendering intencive activities.

-Rick

Re:Not bad...

[LWATCDR]

"What market is it directed towards? I hardly imagine that typical gamers or animators would have the funds for such a card."
NASA, Boeing, the USAF, Lockheed, Dreamworks.... In other words not your typical gamer or animator. I would love one of these on a quad dual core system and a few gigs of ram! Just think of the frame rate I would get in Doom! Just kidding. It is not a typical card and it is not targeted at typical users. The scary thing is it will probably be close to what you get when you buy a $500 Dell in 2010.

My wife gave me two thumbs up...

[Anonymous Coward]

...when I told her that I would buy an ATI card that would allow us to decrease the gas bill for our furnace next winter. Guys, you just have to give your better half a good argument and this graphics card is installed in your computer in no time. Just don't mention that you need to buy a better air conditioner to the summer... she'll discover that one. ;)

Re:My wife gave me two thumbs up...

[Anonymous Coward]

Guys, you just have to give your better half a good argument and this graphics card is installed in your computer

I really don't think most slashdotters cats are going to care what kind of graphics cards they have as long as the computer keeps the cat warm.

Re:My wife gave me two thumbs up...

[eclectro]

Just don't mention that you need to buy a better air conditioner to the summer... she'll discover that one. ;)

And when she does you're either going to ebay that card or be very lonely at night.

Re:My wife gave me two thumbs up...

[dodobh]

At that time, argue that you are going to use the thermal output to generate power for the cooling systems. Bonus, massive uptimes!

Re:My wife gave me two thumbs up...

[Anonymous Coward]

And when she does you're either going to ebay that card or be very lonely at night.

Yeah, but at least you got something to keep warm.

Re:My wife gave me two thumbs up...

[Anonymous Coward]

Actually, electric heating is one of the most inefficient ways to heat a space. You'll pay more in electric running that card to produce the same heat you would burning gas.

Re:My wife gave me two thumbs up...

[ultranova]

Just don't mention that you need to buy a better air conditioner to the summer... she'll discover that one. ;)

Just make a pipe that takes the exhaust of the fans to the outside of the walls. You don't even need to vacuum anymore - the constant wind that now blows through your apartment doesn't let any dust gather.

Re:My wife gave me two thumbs up...

[rvw14]

You haven't seen the price that PG&E charges for gas here in California.

Re:My wife gave me two thumbs up...

[toddestan]

I really don't think most slashdotters cats are going to care what kind of graphics cards they have as long as the computer keeps the cat warm.

My cat would probably approve, though she's still pissed off that I upgraded to a LCD monitor (she used to love to sleep on the top of the old CRT).

Hot hot hot!

[Anthony Boyd]

It's called the FireGL because it puts out heat at levels equivalent to a large fire. -T

Awesome!

[Rank_Tyro]

Now I can upgrade to Windows "Vista."

Re:Awesome!

[spartacus_prime]

Don't waste your time, let Windows Vista crash a lesser graphics card.

Re:Awesome!

[Jozer99]

Appreciate the joke, but for folks out there who think he is serious, Microsoft has said that the Intel GMA 900 and ATI Radeon X200 are the minimum graphics cards for using the "new" DirectX GUI. Vista will work on computers with less graphics systems, but in a compatability mode similar to Windows XP's GUI.

Re:Awesome!

[Anonymous Coward]

... in a year's time.

Re:Awesome!

[weicco]

But you still can't get more than 40 fps in Counter Strike Source.

Re:Awesome!

[somersault]

o_0 what resolution you expecting to be running at? =p

Re:Awesome!

[Professor_UNIX]

Microsoft has said that the Intel GMA 900 and ATI Radeon X200 are the minimum graphics cards for using the "new" DirectX GUI.

I really miss the old days when I knew my friend's EGA card was better than my CGA card and my CGA card was better than a monochrome graphics card. When I got a 386 with a VGA card I made sure to really brag about it to him because he was still stuck in lame 16 color land. Muhahahaha. These days I have no fucking clue if my NVidia GeForce 6600 is slower or faster than an ATI Radeon X200 or an Intel GMA 900. Why can't these companies just come up with a common numbering scheme so that newer and faster graphics cards get higher numbers so we can easily tell why they want us to spend more money on them? And WTF is up with ATI naming it an X200? What the hell does that even mean? Their high end cards used to be 9800 or something. Are they going to roman numerals? Why not just name it the ATI Radeon 10200?

/old and grumpy.

Re:Awesome!

[Low2000]

Didn't they also say that Windows XP worked just fine with 128 megs of ram?

http://www.microsoft.com/windowsxp/pro/upgrading/s ysreqs.mspx

Re:Awesome!

[Cat_Byte]

I'm still trying to get nVidia drivers for Linux to work with frontporch and backportch on my 16:9 tv. They have a really old driver that did that but they broke it. I guess they've been busy working on this firestarter card instead.

Re:Awesome!

[Jozer99]

Ok, heres our new resoltion:

From now on, to be able to tell if one graphics card is better than another, graphics card companies will be required to do the following:

On each new card released, come up with a new connector with more pins on it.

On each new card released, support greater color depth.

I assume this is what you are speaking of.  From now on, my 82398743 pin Radeon X800 will kick your 89437349 pin nVidia 6600!

Re:Awesome!

[Siffy]

I'm already confused... So the lower pin count wins or the higher? Last time I checked, 89437349 > 82398743. Are you already trying to market drastic FUD with your new system?

Re:Awesome!

[toddestan]

Yeah, they need to clarify that a bit. I'm pretty sure that my ATI Radeon 9600Pro 128MB, despite being a couple of years old, is a more powerful card than the X200 which is a budget graphics card. But since the X200 has a higher model number, does that mean my 9600Pro is locked out?

And the Intel GMA900? Good grief. Given the quality of Intel's previous integrated graphics chips, I suspect that my old 32MB GeForce2 has more horsepower.

Re:Awesome!

[Jozer99]

Yes but it is a much more manly connector.

So?

[Tebriel]

Other than high-end graphics work, what the hell will this mean? Are you seriously saying that we will be seeing games needing that must video memory anytime soon? Hell, they have a hard enough time getting people to buy cards with 256 MB of RAM.

Re:So?

[Bitter and Cynical]

Other than high-end graphics work, what the hell will this mean?

Nothing. These cards are not meant for gaming, in fact if you did try and use it for gaming you'd be very upset. The FireGL line is a workstation card meant for things like CAD or Render farms that are very memory intensive and require a high level of precision. Its not meant for delivering a high frame rate and no gamer would stick this card in his machine

Re:So?

[atrus]

The GPU is actually identical except for a few features masked off on the Radeon cards which are available on the FireGL. The framerate would be nearly identical. It would be a monumental waste of money.

Re:So?

[Lesrahpem]

The FireGL series of cards isn't for playing games at all actually. They're geared towards graphically intense things like 3D animation, CAD, and video editing. The same goes for nVidia's Quadro series of cards. Both sorts of cards are geared more towards percision than speed. Your end result are very nice looking images, but lower framerates.

Re:So?

[OldBus]

It might not improve the framerate but there are some games (eg flight simulators - if you consider those to be games) that can produce really nice results if you have very high res textures. I run X-Plane on a 2x256MB nVidia SLI setup and while I get nice framerates, I am often struggling for video memory if I want to run the really highest res textures and objects.

Re:So?

[Alioth]

I had one of the early Fire GL Pro cards (not primarily for gaming). But I also played the odd game, mostly MSFS. I discovered a bug in their drivers that MSFS revealed, but they refused to do anything about it 'because it wasn't a gaming card'. So you'll be doubly disappointed if you find something that doesn't work right in a game because they won't accept the bug.

Re:So?

[Lord_Dweomer]

Are you serious? If you provide the memory, developers will use it. Period. Remember, "256k should be enough for anybody!"

Re:So?

[Threni]

> The framerate would be nearly identical. It would be a monumental waste of
> money.

Wasting huge amounts of money chasing the latest minor increment in the number of triangles plotted per fraction of a second is something a lot of gamers apparantly indulge in. Let them, I say, as it means I can pick them up cheap when they go out of fashion...

Re:So?

[drinkypoo]

I used to have a Fire GL Pro card as well, 16MB I think. It was just slightly slower than a Voodoo 2 card, but it was AGP, a single-card solution, and had dramatically better image quality than the 3dfx offering - which was about the only other thing available at the time.

follow Nvidia into Physics?

[arjovenzia]

With all that beef behind them, i sure hope they will follow Nvidia (i actually have no doubt that they will) in offloading physics to the GPU. http://www.rojakpot.com/showarticle.aspx?artno=303 &pgno=0

it would be nice not having to purchase a top-notch CPU, GPU, and PPU (Physics Processing Unit) in the future, rolling the PPU and GPU together

Re:follow Nvidia into Physics?

[thedletterman]

I'm definately a far bigger fan of Havok's sli physics for nvidia than a 1GB video card. I wish ATI would spend half as much effort not making crossfire such a piece of crap.

Re:follow Nvidia into Physics?

[Cyno]

it would be nice...

Just like it would be nice to get all our computing needs, hardware and software, from one monopoly.

what... teh.....fuk

[Spy Handler]

this graphic card has MORE RAM than my entire computer, and faster too, and a faster processor, and probably a bigger heat sink.

Is it just me or are graphic cards getting ridiculously insane? I know I don't need this thing cus the last game I bought for my comp is 2002 Star Wars: GB. Maybe I'm just a lamer and and l00ser...

Re:what... teh.....fuk

[TheRaven64]

This is a workstation card, not a games card. The people buying this are likely to be either CAD/CAM people with models that are over 512MB (the workstation it plugs into will probably have a minimum of 8GB of RAM), or researches doing GPUPU things. To people in the second category, it's not a graphics card it's a very fast vector co-processor (think SSE/AltiVec, only a lot more so).

Re:what... teh.....fuk

[c_forq]

This isn't a gaming chip, this is part of their pro line. These are for CAD and rendering purposes.

Re:what... teh.....fuk

[temojen]

Stream processor, not vector processor. The programming models are different.

Re:what... teh.....fuk

[dbIII]

the workstation it plugs into will probably have a minimum of 8GB of RAM

Will it really need that much system memory? The first SGI workstation I saw had double the amount of video memory to system memory.

Re:what... teh.....fuk

[Telvin_3d]

Oh, yeah. 8 GB is common.
Part of it is why not? Even if you could get away with 4 or 6 GB, if you are building the type of workstation that would use this chip, bringing the ram up to 8GB is a drop in the bucket as far as money goes. Even EEC 1 and 2 GB sticks of ram is cheap compared to what this card will run.

Re:what... teh.....fuk

[pizpot]

Uhhh, I did CAD for 10 years at an office designing buses. 30,000 parts per bus. We didn't put textures on the parts, and quite frankly that card is extreme overkill for CAD IMHO. Hell, any game card did fine as long as you had lots of RAM. It was a Unigraphics shop.

Re:what... teh.....fuk

[Anonymous Coward]

Yeh!! I was really surprised to see a graphic card with more memory that the ram mounted on the mainboard. Well that was on my 386sx w/2MB vga anyway.

Re:what... teh.....fuk

[Jozer99]

Well, if your computer is slower than this graphics card, either admit you don't need a fast computer or fork over $500 to eMachines to get a modern one. Graphics cards are no clockspeed speed-demons, and this thing is clocked at less than 1Ghz, the speed of a CPU from 2001.

Re:what... teh.....fuk

[t35t0r]

The argument that a workstation card would suck at playing games couldn't be farther from the truth. Have you guys actually tried playing games on quadros and firegl's ?

Re:what... teh.....fuk

[Anonymous Coward]

My voodoo3 2000 has a faster core clock than the pentium cpu in the same box, the voodoo puts out more heat, but there is at least more ram on the motherboard than on the graphics card. Graphics cards have always been insane.

As to why you might want so much ram, I recall an interesting statistic. In developing the game Riven, there was one scene that took over 2 hours just to load into the modeling app they used. There were way too many textures and polys for the workstations of that era.

Re:what... teh.....fuk

[InvaderSkoodge]

It's not just you. It is insane (for the average user at least).

You know what I would like? A million dollars.

Seriously, what I would like is a video card that doesn't make my computer sound like a vacuum cleaner. I've put sound deadening material in my system and installed fan speed controls on all the fans and turned them way down. But none of that mattered because the video card sits there spinning it's fan at like 4000 rpm.

I'm a bit behind the technology curve. My video card is a Radeon 9800. But it works fine for what I do with my computer (not much gaming). Modern cards can do far more than the 9800, so it only logically follows that with this new tech you could make a card that was underclocked and maybe some pipelines turned off, to get it down to the point that a passive heatsink (and not some huge zalman type thing) is sufficient for cooling, and still have better performance than the 9800 that works fine in my system. I'd seriously consider such a video card.

Obligatory...

[Landshark17]

Yes, but will it run Duke Nukem Forever?

Re:Obligatory...

[aurb]

Not only it will run Duke Nukem Forever, it will run Duke Nukem Forever on Windows Vista... Or maybe not...

Re:Obligatory...

[Anonymous Coward]

Hey you idiot, not even the guys at 3D Realms run Duke Nukem Forever!!

Re:Obligatory...

[atari2600]

You mean, it will run Duke Nukem Forever, forever

Re:Obligatory...

[teslar]

Don't you think the card will be obsolete by the time Duke Nukem Forever comes out?

Useful or hype?

[Digital Pizza]

What I wonder is, can current software and hardware make use of those massive specs and is the memory bandwidth high enough for the GPU to benefit from a gig of video ram. or is it all just a gimmick?

Is it worth somone's money to buy such a card?

Re:Useful or hype?

[damsa]

insert joke about Phantom console, PS3 and Duke Nukem here.

Re:Useful or hype?

[Arramol]

Two very valid questions. I remember when the first 512mb cards came out, the benchmarks showed minimal performance gains. Occasionally, they even performed slightly below their 256mb equivalents (though never significantly). Since we're only just beginning to see 512mb cards come of age, this could be history repeating itself. That said, this is a FireGL, not a Radeon, so it's not really designed with gamers in mind. It is possible then that it might actually be worth it provided you're using it for its intended purpose.

Re:Useful or hype?

[Anonymous Coward]

One application of this might be volume rendering. Current medical CAT angiography scans have a resolution of >= 512x512x1024 with 2 bytes per voxels -- too much for a 512MB card considering that you still need some other data on there (lookup tables, etc.).

If you don't have enough GPU memory your only options are a CPU-based approach (slow), custom hardware like the VolumePro board (bloody expensive), or bricking. Bricking means that you divide the volume into sub-cubes and upload & render each of them seperately for every frame which essentially kills performance.

But increasingly powerful cards like this one might make interactive volume rendering possible using off-the-shelf GPUs. Pretty neat!

Re:Useful or hype?

[ChrisMaple]

In CAD work - particularly large area integrated circuits - the more image area that can be cached on the graphics card, the larger the area that can be panned over without re-rendering the image. Re-rendering the image requires re-reading and processing the whole database. It's no fun to try to extend a line beyond the edge currently on-screen only to see the screen go black for 20 minutes while a new image is drawn. Expensive engineering time is wasted in re-rendering, and the engineer becomes less happy with his job.

If it makes a difference, payback is only a few months.

$ 2000 ! yikes.. give me two

[Anonymous Coward]

anyway... this will be half the price by the time Vista ships.

call me again in year 2007

Re:$ 2000 ! yikes.. give me two

[Penguinoflight]

It's a high-end workstation card; the price on these things drops only about 10% for the first 5 years simply because they aren't mass produced, and they aren't mass marketed.

Too bad its still an ATI...

[Jackie_Chan_Fan]

ATI's opengl drivers are flakey on their non firegl line of cards. Some suspect thats by design.

Graphic card makers should get with the program and stop releasing firegl's and quadros. Just release really kick ass 3d accelerators for all.

That way we can all have full opengl support and not the lame opengl game drivers by ATI. Nvidia's gaming card opengl drivers are better than ATIs

Re:Too bad its still an ATI...

[Kagenin]

These cards tend to be for Game and and other 3D Developers, not gamers themselves. Developers need hardware that's about a year ahead of what consumers are currently using. These also tend to be a little more optimized for Windowed 3D apps (CAD/CAM apps, Maya, etc...) than say for Full-Screen apps (games). Consumers don't need that much punch, usually (until the game is released, that is).

I've been using an ATI AIW 9800 for some time. I don't know what you're talking about regarding OpenGL implementation. I'd been using NVIDIA boards up until my AIW (my last was a GF2MX, cheap but punchy, but I'd been using NVIDIA chips since the Riva128), and ATI's implementations are as bad as you might think, after reading your post.

If you wanna cite something, that might help your arguement....

Either way, this card isn't for you or me. But if you're making Maya movies or something, it might be up your alley.

Re:Too bad its still an ATI...

[Baloo Ursidae]

That way we can all have full opengl support and not the lame opengl game drivers by ATI. Nvidia's gaming card opengl drivers are better than ATIs

Not by enough to matter, and will probably never see the developers it needs to work right because nVidia's too much of SGI's bitch to do anything about the situation so they can open-source their drivers...

Re:Too bad its still an ATI...

[somersault]

Wouldnt you rather have a card/drives that have been designed for higher framerates, but with lower rendering precision? You dont need 128 bit floating point precision to play Quake. You're not going to care if a vertex on your health pack is 0.000005whatevers off to the left of where it should be, and you're not even going to see that onscreen unless you go in realllly close. I'm maybe oversimplifying things, but I tried a few years ago running Counter-Strike Source with a Quadro graphics card, then some Radeon that was lying around; the Radeon had much better framerates. It may have been a generation up from the Quadro, but it would still have been way cheaper to buy initially, and people are *not* going to pay for that 'kick ass 3D accelerator' when a more budget version does the same thing, but with less memory, and drivers that are optimised to make games fast, rather than very precisely rendered.

Re:Too bad its still an ATI...

[OneSeventeen]

I'd like to second that!

I still have issues now and then running Blender on my 6 month old ATI card, not to mention my linux laptop having no hardware accelleration because ATI doesn't seem to release too many Radeon Xpress 200M drivers that actually work.

As with all video card battles, I'm assuming nVidia will come out with one any moment now, right before ATI releases a physics engine, and back and forth and back and forth until they both get tired for a while. (all the while I'll be saving up for some previous-generation nVidia card with tried and true linux performance...)

Re:Too bad its still an ATI...

[Jackie_Chan_Fan]

If you wanna cite something, that might help your arguement....

I'm a 3d animator, and a lot of people i know use the high end gaming versions. The quadros firegls and wildcats etc are all nice but their performance gains are short lived when you factor in the price/performance compared to the highest end gaming card version.

You asked for an example of ATI's bad opengl drivers. Softimage XSI and ATI (non firegl) have had some issues with opengl drivers. Things drawing completely odd. For example you select a polygon and it ends up selecting a polygons you didnt click on. ATI has fixed that issue but it took them a good deal of time.

I've done tests with ati and nvidia cards in opengl and the nvidias had faster performance in animation playback in viewports in the major 3 3d apps.

Compare ATI's direct 3D Support in 3dsmax, vs Opengl in 3d max and its a world of difference. Opengl crawls incomparison.

Nvidia doesnt seem to have serious issues with opengl.

Re:Too bad its still an ATI...

[Jackie_Chan_Fan]

Wouldnt you rather have a card/drives that have been designed for higher framerates, but with lower rendering precision?

As a 3d animator, i'd like to have my cake and eat it too :) I'm a graphics whore. I make them and i enjoy them. So i want the best of both worlds.

I'll take speed anyday :) That would include better intergration with 3d animation apps so that all of the transformation/deformation/morphing/expressions/co nstraints could be completely done on the gpu. :)

Oh i just want it all... I'm a technology now, rather than later person.

Re:Too bad its still an ATI...

[somersault]

But what I'm saying it, that you dont notice the extra precision. Just liking 'knowing it's there' seems kinda dumb, if you're not going to use the capabilities. It would be like driving a Landrover/Rangerover in the city (which a lot of people do, yes, and I've had the work Landrover as a company car for a while, but in general it's really dumb :p ). You're not going to need the extra abilities the car has (eg switching to the lower gearing is rather pointless in the city).

I'd take speed as well - at the speeds most games move at, you're definitely not going to notice the odd stray pixel (especially with all the antialiasing and crap that we have these days =p )

Great, but...

[Anonymous Coward]

is there an OpenBSD driver?

ATI/nVidia make gen purpose chips ?

[cyberfunk2]

If these chips are so powerful, and they do seem to be somewhat general purpose (at least by evidence of people making thinks like pi calculators and other small examples utilizing the graphics hardware), why isnt intel/amd using these same techniques with their main chips ?

Re:ATI/nVidia make gen purpose chips ?

[Helios1182]

They are, it is just that things occur on different scales. Graphics processing is hugely parallel, most other code isn't unfortunately. There are fundemental differences that must be dealt with, a general purpose CPU is much more difficult to design than a dedicated GPU.

Re:ATI/nVidia make gen purpose chips ?

[drivekiller]

Looking forward to some really fast GPU-based software for breaking encryption or brute-forcing passwords, then...

Mod this dude up

[Anonymous Coward]

As much as I hate these "mod up" posts, this guy deserves it. It covers everything well. Wish I had mod points to peg this as "Informative"

Re:Mod this dude up

[dotgain]

Wish I had mod points to peg this as "Informative"

Or even "Flamebait" since that's what the blunt, honest truth seems to get these days...

Re:Mod this dude up

[Anonymous Coward]

As much as I hate these "mod up" posts, this guy deserves it. It covers everything well. Wish I had mod points to peg this as "Insightful"

Re:Mod this dude up

[Anonymous Coward]

As much as I hate these "mod up" posts, this guy deserves it. It covers everything well. Wish I had mod points to peg this as "Redundant"

Re:Mod this dude up

[Anonymous Coward]

As much as I hate these "mod up" posts, this guy deserves it. It covers everything well. Wish I had mod points to peg this as "Recursive"

ATI Sucks for driver support with Linux

[Vskye]

In a nutshell, see the subject.
I really don't give a flying *uck if any company, be it ATI or Nvidia comes out with the latest and greatest video card if it does not have proper driver support! Anyone who's run linux for awhile knows the drill.

Re:ATI Sucks for driver support with Linux

[kitejumping]

thats why I buy nvidia... if the performance is the same, may as well have driver support.

Re:ATI Sucks for driver support with Linux

[barefootgenius]

It does seem that they cheat somewhere doesn't it. Tonight I'm putting my 9600 pro into a friends computer because my old Geforce 4 works just as well. It doesn't get the framerates that the Radeon does but its much smoother and the drivers aren't a bitch to install.

On the other hand, at least ATI does Linux drivers.

Re:ATI Sucks for driver support with Linux

[Anonymous Coward]

I agree. Why is ATI flooding the market with all these video cards without any concern towards drivers for alernate operating systems. This crap reminds me of the days of the Mach8 based cards with no drivers for OS/2. Get in the game ATI, your name in non-Windows circles represents crap. Where are the Linux drivers for 1X00 based cards that have been on the market for months now.

G.

ATI/nVidia make special purpose chips ?

[Anonymous Coward]

"why isnt intel/amd using these same techniques with their main chips ?"

MMX,SSE,Altivec. Of course one could ask why isn't SATA, or sound integrated into main processors, or any other specialized task? The reason is that the GP in GPU doesn't stand for general purpose. Despite the work at GPGPU, the GPU's domain is narrower than what a CPU addresses.

O.O~!

[KajiCo]

...

If only I didn't have to sell my car to buy one :/

Whoa.

[RoffleTheWaffle]

Okay, if this thing is seven times more powerful than a three-gigahertz Pentium processor, why the hell aren't these guys making CPUs alongside their GPUs? Seriously, GPUs have gotten to the point at which they are just as if not more powerful than standard CPUs, and with quantities of RAM to match a whole PC. The news of this new and extremely powerful GPU - should it stand up to the hype - alongside news that NVidia has developed a physics coprocessing technique using SLI leads me to believe that GPUs may soon no longer be just GPUs, but complete co-computing units to which graphics, physics, and other demanding tasks could be offloaded. That might be interesting.

Re:Whoa.

[atari2600]

Here's an analogy:

Company A: Company A makes hand held projectile weapons (insert your favorite weapon here). Their latest and greatest hand-held, one man operable compact weapon is twice as powerful as an Anti-aircraft gun (can punch a 2x bigger hole in a cuboid of lead).

Company B: This company makes anti-aircraft guns that, well, have one purpose.



Kid who knows nothing about the Instruction set and complexity of the process: Why THE HELL doesn't Company A make Anti-Aircraft guns since they make GUNS THAT ARE TWICE AS POWERFUL (when it comes to shooting through lead).

Think about it. (Also read the post near the top of this page/post).

Re:Whoa.

[Jozer99]

Explained in detail above. Suffice it to say that CPUs and GPUs are radically different. With GPU's, ATI can throw out old architectures and create new ones whenever they want (quite often). Since the hardware is accessed by a driver, the user isn't limited in what programs they can use. With CPU's, everyone is stuck with x86, which was invented in the 1980s. You can't break compatability with x86. GPUs do mostly simple floating point calcuations. Therefore, they are basically massively parallel FPUs. If they need to do a non-floating point calculation, they are quite slow. CPUs can do floating point calculations, but also many other types of calculations, and are about equally good at everything. For the sake of heat, power consumption, size, and cost, the FPU on a CPU is not nearly as large as a GPU. If each processing unit on a CPU was the size/power of a specialized processor (GPU, ect...), the chip would be gigantic, and so would be hard to make, expensive to buy, consume massive amounts of power, and emit unimaginable heat.

Re:Whoa.

[Anonymous Coward]

With CPU's, everyone is stuck with x86, which was invented in the 1980s.

Technically the 1970s, unless you mean the 32-bit extension.

Re:Whoa.

[Anonymous Coward]

"...With CPU's, everyone is stuck with x86, which was invented in the 1980s."

Unless you run a different processor like a PPC. That's why I use a Macinto... whoa - wait a minute...

Re:Whoa.

[Ozymand E. Us]

Would be hard to make, expensive to buy, consume massive amounts of power, and emit unimaginable heat.

So...Intel?

*Ducks*

Re:Whoa.

[Burning1]

If each processing unit on a CPU was the size/power of a specialized processor (GPU, ect...), the chip would be gigantic, and so would be hard to make, expensive to buy, consume massive amounts of power, and emit unimaginable heat.

If you showed a modern CPU to a guy from 1996 he'd say exactly the same thing. He'd probably also drool onto your motherboard.

New for 2010!

[FlyByPC]

A fire-breathing, liquid-helium-cooled 256-core graphics engine, complete with 2KW independent power supply, temperature throttling, and VR interfaces.

(Oh, yeah -- and we think there may be a CPU in there somewhere, too.)

damn.

[popeguilty]

Now I have to change my pants.

ATI's Linux Zealotry Support Sucks

[Anonymous Coward]

"I really don't give a flying *uck if any company, be it ATI or Nvidia comes out with the latest and greatest video card if it does not have proper driver support! Anyone who's run linux for awhile knows the drill."

Gee. Zealots don't give a flying fuck about non-zealots. Film at eleven. The people who will be using this card aren't going to be using Linux. Therefore your don't care attitude is rather irrelevent. For the domain were Linux will be used. Video cards aren't being used. e.g. clusters. So once again your don't care attitude is irrelevent.

"workstation cards" what r they 4?

[2ms]

I use Pro/E as my job right now but have only ever used it on this 4 year old Dell "workstation" with some kind of 4 y/o Fire card in it. But though this equipment is old, I have never for a second felt like anything was slow at all while using Pro/E. So what are these insane cards for? Im sure they are for something I just would genuinely like to know what it is. I would have thought they were for CAD but now I see other people running Pro/E just fine on like laptops with integrated Intel graphics.

Re:"workstation cards" what r they 4?

[Mithrandir]

I work a lot with the visualisation end of the market and recently have been working with NASA on the CEV project(s). Some models that we deal with are in the gigabyte file size just for the geometry for a single subassembly. This card would make viewing some of these things far easier as you can preprocess and schlepp almost all the geometry to the video card as a VBO and never have to pass it over the bus again. Makes for tremendous performance gains.

Re:"workstation cards" what r they 4?

[LookoutforChris]

As a fellow Pro/ENGINEER user this is not my experience. What version are you using and how big are your models? The latest version is a hog (as always). I can't imagine using it on an old Dell with a FireGL and doing anything very complicated. I have to admit I'm not a fan of ATI cards, their OpenGL support seems to be very flaky. But I like the larger memory on these new cards and the price is good. Price wise this card would seem to compare favorably to a top model WildCat Realizm or a top model nVidia Quadro.

Re:"workstation cards" what r they 4?

[bhima]

Funny you should word it like that...

I just finished a meeting with our lead Mechanical Engineer who uses some dual PA-RISC thing from HP, which if memory serves, is around 2 or 3 years old. His comment (while showing use various models of the current project) was that if the next project increased as much in complexity as the current project did from the former he would like to get new workstations for his team.
I should also comment that it appears to me that given the right people Pro/E can become surprisingly complex quickly... for example from the way I understand it we have the equivalent of CVS running for all parts and components modeled in Pro/E.

I guess I shouldn't complain because my compile times are still less than the time it takes to get a cup of coffee.

As a side note... having seen the various demonstrations of multi-touch HDTV tables I have to wonder what the requirements for a fast & slick UI will be in 2007~8

Not likely OS level

[HornWumpus]

But you can write very specific GPU code to solve some parallel FP problems.

For more general purpose FPOPs you will have a hell of a time getting enough gain in floating point performance to overcome the overhead chatter between the CPU and GPU that would be required to keep the states in synch.

I'd go so far as to say if the process can't be near complely moved (the CPU will need to feed it data and suck up results) onto the GPU then don't bother.

But I'm talking out of my ass, it could work. I'm just skeptical.

Saweet...GIS time

[The GIS Guy]

Now to light a fire under the IT dept's rear to me one of them

next thing you know...

[Anonymous Coward]

ATI might just add a CPU on the card in order to boost gaming performance and delivery the best gaming experience

Re:next thing you know...

[Ginger Unicorn]

I was just thinking that. You never know one day AMD might eat NVidia or ATI or vice versa and we can get multicore CPUs with GPUs built in. That would be pretty bitchin'. Thinking about it, it's kind of like the old 386/387 days when you had a seperate CPU and FPU. Then the DX came out and combined the two. Can't wait till i get an Athlon128 FX9900GT with 2GB cache(!)

Re:next thing you know...

[somersault]

heh, there goes your upgradability.. either you'd have to be able to swap out the chips, or just buy a whole new CPU/GPU (which sounds expensive) each time. But it does sound very cool, and the performance gains would likely be worth the price.

Re:next thing you know...

[NetRAVEN5000]

"You never know one day AMD might eat NVidia or ATI or vice versa and we can get multicore CPUs with GPUs built in. That would be pretty bitchin'."

Sweet! Then when I want a faster GPU I can shell out some more cash for a newer CPU and motherboard that I don't want!

Re:next thing you know...

[Ginger Unicorn]

Yeah cos every time i upgrade my CPU i really resent having to buy a new FPU for it at the same time. Why oh why did intel invent the DX? It all been down hill from there.....

And don't get me started on L1 cache..... why cant they just leave the cache on the motherboard, so i dont have to buy more of it when i upgrade my CPU. And RAM on my video card! What the hell is that? Why can't it just use system RAM?

Oh wait, because its much faster and and more efficient and my Athlon with it's built in FPU and on-die Cache is not more expensive than a 386SX+387 was in it's day.

Re:next thing you know...

[NetRAVEN5000]

Which is a perfect explanation why I should have to upgrade my CPU when I want a new GPU. Yup.

Think about what you're suggesting. First off, look at Intel's GPUs. They don't even compete with nVidia's and ATi's GPUs! Do you really want to be stuck with a mediocre GPU just because you don't want to shell out the cash for a new CPU? Many high-end CPUs already cost upwards of $500, as do many high-end video cards. Why force people to spend $1000 for a new CPU when all they really want is a new GPU? It'll turn away many potential buyers.

Not to mention, where are you going to put the video RAM?

Or the fact that you're taking things out of proportion. The CPU and the FPU are almost directly related in that they both do math calculations, so they should be together. The GPU and the Video RAM are both used for video, so they should go together. But the GPU and the CPU - one does graphics processing, one does a multitude of different calculations. They shouldn't be put together. When you want better graphics, you usually upgrade your GPU - not your CPU. When you want your computer to be more responsive overall, you upgrade your CPU - not your GPU. What makes you think it's a good idea to force people to upgrade both when they only need/want to upgrade one?

Re:next thing you know...

[Ginger Unicorn]

You're thinking about it in terms of what things cost now. What i'm suggesting is that in the future it is possible that CPUs will have on-die GPUs and have large enough cache to accomodate general and graphics activities. And of course all this would be at the usual CPU price range.

What would effectively have happened is the seperate GPUs would be obsoleted or relegated to highend workstation parts, since you could get an all-in-one for $50-$500 depending on spec. This $1000 you're assuming it must cost is based on today's technology. This idea of being "forced" to upgrade is akin to being "forced" to upgrade your FPU every time you buy a new CPU; Since it would be inclusive in the price, you just wouldnt think or care about it.

This really isnt that fantastical to imagine. It's far from impossible, and the effieciency gains suggest to me that it would a probable development.

As for the arguement that they are designed to do seperate things- the reason this integration idea occured to me in the first place is that earlier in the forum someone pointed out the GPU is just a massively parallel FPU and that they are becoming more and more generalised, and are now being used for other FPU tasks, eg. Physics. So why not replace the bog standard FPU in a CPU with a massively parallel one and eliminate the need for a seperate graphics board?

Just because Intel make baseline GPUs doesnt mean AMD and Nvidia or ATI could team up and make high performance integerated ones.

Re:next thing you know...

[NetRAVEN5000]

I disagree.

Here's what I'm getting at: I upgrade ALL my other components more frequently than I upgrade my CPU. In fact, I never upgrade my CPU unless the other one either crapped out or is more than 3-4 years old. My GPU, however, I upgrade more frequently - I usually get a new video card or more memory whenever I get gift cards for Best Buy or wherever. Why? Simple - for games, the CPU speed doesn't much matter so long as it's above 1.8GHz. And that's mainly because of all the physics that is programmed into today's games - once they start making physics accelerators (I remember reading about some startup planning to make one) you won't even need it that fast.

The performance of my GPU is what I care about. I don't want to buy a new CPU. I don't care if it only costs the price of a new CPU because I don't want the CPU.

You're right, CPU manufacturers could team up with GPU manufacturers to make a nice high performance CPU-GPU combo. But then what happens when ATi or nVidia comes out with a new GPU? The only way to get the new GPU will be to buy a new CPU - but if your old CPU suits you just fine, what can you do? Not to mention that you won't be able to use the Intel chips in an AMD motherboard - what if you want to switch from Intel to AMD? Right now I can just pop my GeForce right in any AGP-enabled motherboard and go.

Also, what if you're partial towards certain companies? Some people prefer AMD, some prefer Intel. And some people prefer nVidia, some prefer ATi. What if you want an AMD processor and an nVidia GPU, but nVidia struck a deal with Intel and only makes their GPUs for Intel processors?

Re:next thing you know...

[Ginger Unicorn]

But if a CPU/GPU combo was the same price as a GPU on it's own or CPU on its own why would it be an issue that you're getting a new CPU into the bargain?

I'm not imagining a case where a CPU/GPU combo is some kind of premium option, I'm thinking more like seperate GPUs would become obsolete since their functionality would become part of a standard CPU. And since GPUs were designed in the first place as a kind superpowered mutant FPU, it only seems natural that once the technology gets to the stage where such an FPU is viable as part of the main CPU it would be a lot more efficient to build it there. GPUs seem to be becoming more generalised, gradually moving the graphics specific logic from the hardware to the drivers, and leaving the hardware purely to hammer the neccesary floating point ops through.

I understand what you're saying about not wanting to be forced to uneccesarily upgrade components, but think of it like this, when you buy a new motherboard does it put your nose out of joint that you are getting a new built in soundcard and ethernet card as well? The thing is it's no cheaper to buy a motherboard without those things on it, so what the hell.

Obviously today a CPU+GPU would be an absurdly expensive device, but everything is to start off with, i'm talking about when it isn't expensive anymore.

And i'm not just thinking lump these things in together for the sake of it. The biggest bottleneck with GPUs today is latency. An on-die GPU that uses on-die cache is going to be phenominally fast.

As for brand loyalty, who knows, maybe ati or nvidia could cook up their own CPUs. Or Intel / AMD could add in their own GPUs. Kind of like a more useful SSE.

Re:next thing you know...

[NetRAVEN5000]

My problem is, what GPU do you get? I want a good, fast GPU whereas a business user might not even want one at all. Likewise, he might want a nice quick CPU where I don't really care too much. Sure, one day maybe it'll cost just as much to buy them separate as it will to buy them together, but if I can pay less to just not get a new CPU, I'll opt for that option.

As for with getting a motherboard with extra features on it. . . it depends on the price of the motherboard. I just bought a new CPU/motherboard combo for ~$160 (my other one fried two PSUs in about 3 weeks so that's why I got a new one). The new motherboard has great sound, Gigabit Ethernet, and plenty of USB ports - it was a great deal! But if it were more expensive, I certainly would've done without Gigabit Ethernet - my house can't even handle more than 100Mb/s anyways. I would've just used my old Ethernet card.

Now if they could come up with a way to make a CPU with a better GPU for cheaper than (or the same price as) the GPU alone, I'd buy it. But I doubt that'll ever happen. Or maybe I'd buy the CPU-GPU combo when I'm rich.

Now, maybe if there were a spot directly on the motherboard to attach a GPU (meaning not just an AGP slot for an actual card, but a spot for an actual chip). . .

ATI works great for me with Linux

[deek]

Flogging generic statements like "ATI sucks for Linux", is not very accurate. A better way of putting it is "ATI sucks for some cards under Linux".

I can certainly say that my laptop, with its ATI Radeon Xpress 200M chip, works wonderfully under Linux. Yes, I'm talking about their binary driver distribution. Using the latest version of their drivers. I'm also using the Xorg 6.9 xserver. It's fully 3D accelerated, as shown in the following command:

$ glxinfo | grep OpenGL
OpenGL vendor string: ATI Technologies Inc.
OpenGL renderer string: RADEON XPRESS 200M Series SW TCL Generic
OpenGL version string: 2.0.5695 (8.23.7)

I'm aware that the binary driver doesn't work with some ATI cards, especially some of the top range ones. But for what I use, it's brilliant. The installer is a little easier than the Nvidia one too. Thanks ATI, you've done a great job, from my perspective.

Re:ATI works great for me with Linux

[yellowcord]

Try turning on the whiz bang hardware accelerated effects in KDE. You get a nice garbage screen if you turn on "Use translucency/shadows" in Window Behavior under desktop in the Control Center. Apparently it works great on the FOSS drivers for the Radeon 8500 and below but I get a screen full of garbage on my 9800. Probably not a big deal for most people but its very annoying to me.

Also frustrating is the lack of support for games played through Cedega, I just installed the Cedega timedemo tonight and Guild Wars almost but not quite works. I have characters missing heads and the frame rate is about 2 frames per second. I can get decent frame rates with a different setting but it crashes if I go to another district. This may be fixable, but so far my next computer is going to have an nVidia GPU.

Re:ATI works great for me with Linux

[Grey_14]

As someone who's been getting increasingly into OpenGL programming, I can tell you GlxInfo isn't everything, just because it say's it supports opengl doesn't mean it won't randomly segfault when you throw a supposedly supported feature at it :(

Re:ATI works great for me with Linux

[Mo6eB]

Yes, ATI tends to cheat quite a lot in that regard. The particularly hilraious bit is where it says that they support OpenGL 2, while their drivers are still missing some functionality from 1.5. A while ago, I noticed, that when you query for supported OpenGL extensions, the GL_ARB_draw_buffers extension gets reported as supported, but any attempts to dynamically load glDrawBuffersARB fail. So I submitted a bug report, and now, 2 driver versions after that, it is still unfixed (correct me if wrong). Quite staggering, considering the fact, that there is a glDrawBuffersATI. One would think that some person would put a glDrawBuffersARB, that calls the ATI version, but it seems that they are too lazy for that.

Re:ATI works great for me with Linux

[lubricated]

I have a similar chip in my laptop. Xgl causes the system to completely hang itself. That is not a good driver.

so...

[Anubis350]

Are you saying it renders flames *very* realistically? :-P

Re:so...

[Baloo Ursidae]

Heheh, SternoGL...

Re:so...

[dodobh]

Implying that this is a good card for use when reading usenet, slashdot and talking on IRC?

Re:so...

[Beriaru]

Connected to a Phillips Ambilight is umbeliable!

Re:so...

[Anubis350]

Yes. For the wars and for burning karma :-P

Re:so...

[pcgabe]

Denholm: Nice screensaver... I love the way the smoke seems to be coming off the top of it. (The IT Crowd)

Re:so...

[somersault]

wow, I'll get some of that stuff slapped between my CPU and heatsink right away! :D

Can i reallocate that memory as system memory?

[mOOzilla]

Can i reallocate that memory as system memory?

Re:Can i reallocate that memory as system memory?

[ForestGrump]

Humm, considering everything is now PCIe, with 2.5 gigs of bandwidth either way it doesn't look too bad- considering the therotecial bandwidth of ram in most machines is in the 2-4 gig range anyway.

So if there isn't much happening on the graphics end of things, allocate some ram on the graphics board as swap space for the machine. However, when you fire up a game or something, have fun waiting while everything in graphics ram is swapped to hdd.

Grump

Re:Can i reallocate that memory as system memory?

[junics]

I remember reading something about that in an old article, but paging/swapping or a ramdisk would be the only practical use. http://hedera.linuxnews.pl/_news/2002/09/03/_long/ 1445.html

Re:Can i reallocate that memory as system memory?

[Stephen Williams]

On Linux systems, it's possible to use video RAM as swap space. Here's an article explaining how. The article is a few years old, and I can't vouch for it personally because I've never tried it, but it might at least provide you with a starting point.

-Stephen

Width of the floats?

[mosel-saar-ruwer]

Signify: full 128-bit precision

TheRaven64: or researches doing GPUPU things. To people in the second category, it's not a graphics card it's a very fast vector co-processor (think SSE/AltiVec, only a lot more so)

Traditionally, ATi floating point numbers were only 24-bits wide [i.e. only "three-quarters" of single precision, which is 32-bits].

nVidia, IBM Sony Cell, and Altivec support only 32-bit floats.

MMX supported no floats whatsoever. SSE supported 32-bit floats. SSE2 & SSE3 support 64-bit floats. x86 supports 80-bit floats.

So what is this 128-bit stuff all about?

I don't suppose there's a chance in hell that these could be quad-precision floats, could they?

Re:Width of the floats?

[temojen]

Most likely 4d single precision vectors or 2d double precision vectors if they want to market it to researchers who had been cooking their own stream processors from FPGAs

Re:Width of the floats?

[antime]

nVidia, IBM Sony Cell, and Altivec support only 32-bit floats.

The Cell SPEs do support double-precision floats, but IIRC you take a massive performance hit when using them. The PPE has a normal floating-point unit in addition to the Altivec instruction set.

Re:Width of the floats?

[Anonymous Coward]

Yes, they are 128-bit floats. They're needed for doing HDR.

Ars Technica discussion on why so much GPU memory

[Brett Johnson]

Last year, Ars Technica talked about how newer OS's are leveraging fast GPUs for advanced graphics. The main problem is the bottleneck between system memory and GPU/VRAM. One solution is to move the bottleneck to the other side of the backing store.

http://arstechnica.com/reviews/os/macosx-10.4.ars/ 13
http://arstechnica.com/reviews/os/macosx-10.4.ars/ 14
http://arstechnica.com/reviews/os/macosx-10.4.ars/ 15

now if only they knew how to make drivers

[xamomike]

sounds great and all, but have they gotten around to paying their own programmers to make drivers that actually work, and install off the CD it comes with, instead of outsourcing it to a few guys in their basement?

Seriously, I've owned 6 different ATI cards of differing lines this year, and only 2 of them installed properly with the drivers that came on the CD. That just aint right.

Re:now if only they knew how to make drivers

[ericartman]

Agreed Screaming hardware, no drivers. must be an ATI product again. I'll never buy another one!

Re:now if only they knew how to make drivers

[manno]

Are you crazy you actualy use the drivers supplied on CD's? I thought everyone knew that those are usualy outdated piles of junk, be it nVidia, ATI, Via, SIS, Intel, HP, Creative, Ricoh... anybody. Always get the latest drivers off the website and toss the CD's that come with your hardware, immediatly if not sooner.

Yes but

[Anonymous Coward]

Does it get you banned in World of Warcraft?

No

[temojen]

But it will run Duke Nukem really smoothly until enough silicon decays into phosphorus to make it stop running, or the owner gets tired of Duke Nukem, whichever comes first.

Re:No

[Baloo Ursidae]

I'm fairly convinced that waiting for the next Duke Nukem is more entertaining than actually playing the next Duke Nukem. If there's a next Duke Nukem. I'm not convinced it's not just a rumor mill scam to allow them to keep pressing the venture capital button eternally...

256 MB is small

[emarkp]

Try rendering medical image data as a 3D texture (well three textures actually, one for each primary image). With 300 images, 256KB per image, x3 textures, that comes out to 225MB just for the textures. I deal with datasets like these routinely, and more video memory is a welcome development.

you don't render pixels independently...

[YesIAmAScript]

For example, if you need to render a triangle, transforming the coordinates and finding the u,v (for texture mapping), then calcuating the normals and interpolating them all has a lot of setup math. You don't want to do this setup math for each pixel.

In fact, unlike ray tracing, where you proceed pixel by pixel, checking all the geometry, in modern techniques you render the parts of the geometry in order into the frame buffer, and due to the magic of z-buffering, the end result frame buffer is complete for all pixels at the same time.

So pretty much what you say about pixels being independent is nearly completely false.

You can still parallelize a lot of stuff, but it isn't as simple as you say and is done using different techniques.

Re:you don't render pixels independently...

[vux984]

the end result frame buffer is complete for all pixels at the same time.

Actually, that's pretty much the definition of a parallel process. ;)

The point is the completion of one pixel isn't dependant on the completion of others. The fact that so much of the 'setup' is the same for all pixels just means you can acheive the equivalent of a million independantly computed pixels with far less actual effort.

In other words, if they -weren't- so easy to parallelize these optimizations would be impossible to make.

Re:you don't render pixels independently...

[Anonymous Coward]

So the dependency chain is at least two levels deep. Horror!
Contrast with e.g. ciphers in CBC mode where the chain is roughly as long as the input.

Not for gaming, for graphics workstations!!

[Anonymous Coward]

Ok, I cannot beleive the absurd number of posts I am seeing from lamers who think this thing is for video games. Hello People! Both ATI and NVidia have had seperate high-end workstation lines for years now! This is nothing new. Where have you people been?

This card is for people who need serious rendering of high detailed scenes and 3D objects, not serious frame rates for games. For applications where image quality, complexity, and accuracy are much more important than frame rate. The GPUs in these high end workstation cards are geared in a totaly different manner and actually suck for video games! These are great for CAD/CAM, medical imaging (like from CAT and EBT scanners), chemical modeling, and lots of other hard core scientific and 3D developement type stuff.

Re:Not for gaming, for graphics workstations!!

[nehumanuscrede]

" This card is for people who need serious rendering of high detailed scenes and 3D objects "

Actually no. I'm guessing though that you meant ' display ' instead of rendering.

The graphics card has zero to do with the final rendering of any 3D scene. THAT responsibility falls to your processor(s) and Ram.

The gfx cards power and ram capability allow you to manipulate the object more efficiently. Especially as the poly counts on the aforementioned object / scene start to get silly. Start talking about a scene with several MILLION polys and you start to see where workstation cards come into play. Workstation cards are designed to be rock solid stable. They are not designed to do 8 bazillion FPS in Unreal Tournament. . . :)

Many of the 3D apps out there ( I personally use Softimage|XSI ) are capable of OpenGL ' textured ' views of the object you're manipulating. This is where that gfx card ram comes into play. If you want an idea of what your final scene is going to look like without having to use a test render, you use the textured mode to view all of your scene models. High resolution textures eat up a LOT of memory, thus 1GB of onboard ram isn't really that far off the mark for what would be nice to use.

Unfortunately these cards usually cost an arm and a leg. Bottom line ? Not even a consideration nor even worth looking at unless you're a 3D professional with studio funds backing you up :)

I would expect to see a counter from Nvidia in one of their Quaddro model lines. . . .

TV-out anyone?

[giorgosts]

Does any one know how to make TV-out (PAL) for Radeon 7500 work? I've used everything, atitvout, different distros, fglx, open radeon, all. The best performance I' ve seen so far is with Ubuntu 5.10 on the desktop, but all I can get from X on the TV is an incomprehensible image with the same colors as on the monitor.

Re:TV-out anyone?

[Anonymous Coward]

I configured a mode in my xorg.conf file with 800x600 and 60Hz refresh. To enable tv output all I have to do is make sure that my TV is on when I boot my computer (once a month) so the video BIOS will enable the output. I did this on FreeBSD, should work fine on Linux too.

Re:TV-out anyone?

[giorgosts]

Monitor is 1024x768 @85Hz. Fine with monitor (doesn't flicker), TV-out works with linux kernel (what I see on the monitor I see on TV). When X boots just an incomrehensible picture. I got the same with all Live CD distros I tried, even with a Radeon 9600 XT AGP I borrowed from a friend. Now I have to burn the movie I download every day and watch it on XP cause there arent any xp drivers for LVM ext3 also. I will gladly pay for a new card or linux software so I just dont burn and dual boot every time I want to watch tv from my computer.

Finally!

[LookoutforChris]

That's 1GB of unified memory, so less than 1GB is available for textures ; (

It took them long enough; this is definitely the direction to go.

Almost 4 years ago Silicon Graphics gave a final revision hurrah to their best graphics product: InfiniteReality. A pipe sported 1GB dedicated texture memory, 10GB of frame buffer memory, 8 channels per pipe, and 192GB/s internal memory bandwidth.

And an Onyx system could have up to 16 pipes! That's 8.3M pixels per pipe, or 133M pixels from a full system! And all in 48-bit RGBA. And those are just the raw numbers, there were a great many high end features only found on InfiniteReality. Don't ask what it costs ; )

Sorry for the passionate post. It seems that Slashdot is very PC-ish and narrow in its viewpoint (Imagine a Beouwolf of... Can it run Doom3 ... etc.) so I couldn't resist blabbing about high-end kit that's off topic.

I've had the pleasure of using a small Onyx system. Too bad SGI is dead dead dead. Still they provide a good target for everyone to shoot for. Some day the above power will be available for a few hundred dollars for the average person. Though I think it will be atleast 5 years before the quality and features of InfiniteReality4 are at a consumer level. And never will we have workstations like SGI's again ; (

Re:Finally!

[master_p]

Why does the PC only have 8 GB/sec (theoritical maximum on most advanced systems) bandwidth between CPU and main memory? PCs could have 192 GB/sec bandwidth, but Intel and other high tech companies want to maximize their profits by selling new technology one piece at a time.

Re:Finally!

[LookoutforChris]

I just looked up spare parts in the SGI price book. It's over $90,000 for a basic InfiniteReality4 pipe, about $250,000 for a fully configured pipe (4RMs and a DG5-8), or $3.25M for a full 16 pipe system (including KTOWN2 boards). I'm not sure if buying the graphics as part of an original system saved you money or not.

I'm sure that $$$ the main reason you don't see these in PCs.

Re:Finally!

[master_p]

Not really. 10 years ago an NVIDIA 7800 GT would cost 10,000$. Next year it would cost 100$.

Yeah. I'm applying for a patent

[toadlife]

It all started with the GeForce 6800 GPU a couple of years back. My patented process provides a "double-whammy" argument for the wife - A lower heating bill + a lower electricity bill!

You can check out one of the diagrams I provided in my patent application here.

Thankfully

[SuperKendall]

Are you saying it renders flames *very* realistically? :-

Yes but thankfully due to todays powerful use of offloading work to support cards, you'll still be able to hear the fiddles from the sound card until the last possible moment.

And the answer is

[Anonymous Coward]

Absolutely noone cares!

oooo...

[temojen]

Imagine ImageMagick ported to that... No really, I've been waiting ~15 minutes for the output of identify -verbose on a 25MPixel 48bit RGB image. I'm sure glad I didn't use the 100MPixel scan.

Re:oooo...

[dwater]

not that graphics power has any effect on ImageMagick...so one of the above systems in it's Origin (ie no-graphics) config would work just as well - they can have many many CPUs (though they are each distinctly underpowered by todays' standards) - I wonder if ImageMagick is multi-threaded...if you have more than one image, then fine :)

Re:oooo...

[temojen]

not that graphics power has any effect on ImageMagick...

But it could, as most of it's algorythms are highly parallelizeable, and modern graphics cards can be used as stream processors.

Re:oooo...

[dwater]

yes, could, but doesn't.

I also note that SGI's old O2 could also have 1GB of memory, as well as the 320 (upto 1GB) and the 540 (upto 2GB), although they all shared that memory with the rest of the system in some way...

Does it run Linux?

[NadaTech]

no, not IN Linux, I mean with 1GB, it should be able to RUN Linux in a VM or something. Ah, OS communities next project - VRAM based OS!

Open-Souce Graphics Card

[Anonymous Coward]

Wasn't somebody planning to design an Open-Souce graphics card a few years back? If so, what happened to it?

ATI loves you in the ass

[ThoreauHD]

http://wiki.duskglow.com/tiki-index.php?page=Open- Graphics
http://www.osnews.com/story.php?news_id=13844

If you are talking about XGI, ATI just bought them and closed the code.

http://www.linuxgames.com/news/feedback.php?identi ferID=8255&action=flatview

Hard ??? - Exceeding Intel FP Performance

[riscguru]

Ever heard of Power5 processors for AIX? Not hard to beat FP performance of Intel processors.

holodeck

[PermanentMarker]

That's a lot, my first pc had16mb video ram. So this device is capeable of 10024/16 ... wow.. 64 monitors ? what to do with that amount of screens..thinking.. Ehm well what did i run in those days ehm WP5.1 Having text all around me would be verry much like my desk and would resamble my desk. So that's enough for a verrry first holodeck !!

Re:holodeck

[Tanamo]

16MB? 16MB? I would've killed for 16MB. Young people today, don't know they're born, etc, etc...

Re:holodeck

[PermanentMarker]

:) actualy thinking again my first PC had less video memmory. So i can increase the number of screens and lower my holodeck application to edlin This idea idea turns a holodeck into nightmare, as i forgot most of edlins commands. (altough i'm not sure either if Shift f7 would have got me to wp5.1's print menu, might have been shift F5 too.)

Maybe not...

[heybiff]

...but I'm sure it'll come with a Fast(TM).

Heybiff

This is great news to software developers

[Anonymous Coward]

Now I can program my graphics engine even more poorly, and when people complain it runs slow, I'll just tell them to buy an even better graphics card!

Damn, you're young.

[Anonymous Coward]

16mb of video mem, that it? Sheesh. You're a youngen. o.O

Who writes this stuff?

[necro81]

FTFA:

The result should be a finer level of detail throughout the visible spectrum, enhancing details in shadows and making highlights come to life.

Unless I'm the Predator and have some special monitor that no one else has, this comment about the "visible spectrum" is ridiculous. Of course it's going to improve fidelity throughout the visible spectrum, do you think they'd just focus on the color green, or try to improve that all-so-important infrared fidelity?

Based on a cutting-edge 90nm process technology

90 nm is not the cutting-edge, it is the industry standard for most performance processors. Even that statement is weak, since the newest Intel offerings are based on 65 nm processes.

combined with full 128-bit precision

As opposed to partial 128-bit processing?

Was this article merely quoted verbatim from ATI's press release? It is filled with so much superfluous hype that getting to the substance of it is hardly worth the effort. Stop after the first two paragraphs and save us three minutes of time.

Good thing this is a workstation card

[billcopc]

It's a damned good thing this is a workstation card. I've never really understood the true purpose of these beasts, and how they are supposed to differ from gamer-class graphics cards, but I'd hate to see what kind of idiot mayhem would ensue if ATI released a 1gb gamer card to the public.

I remember when they started having 256mb budget cards and I kind of scratched my head over the concept of a card that sports as much RAM as its host PC (in many cases), and more importantly why the hell would anyone want 256mb of textures on an already stuttering GPU.

Sure enough, the imbeciles would bowl into the store, fresh off the short bus and ask for the 256mb card, with the religious belief that it was "better" than the 128mb version. What's even funnier is the concept of "TurboCache", in which a card will advertise 256mb of display memory, yet only have 64mb onboard - it borrows the balance from your system ram as needed. Funny, back in the good old days we called it AGP GART.

I'm sure if ATI Released a 1gb Radeon 9200, a whole nation of ignorants would snap it up and think it's a significant upgrade. Hell, Future Shop (Best Buy) would probably dedicate a whole freakin' aisle to stocking them because it would be their top seller. Ironically, the recent leaps in graphical detail have not spawned from insane texturing, they've been driven by elaborate pixel shaders, because really the biggest difference between a 256mb budget card, and a 256mb 7900GTX is the number of simultaneous shader ops, followed by clock speeds.

Re:Good thing this is a workstation card

[argent]

why the hell would anyone want 256mb of textures on an already stuttering GPU.

Because 90% of programming is an excersize in caching, and if you can just cache the textures you can let your GPU just get 'em instead of waiting for it to finish saying "g-g-g-give me a t-t-t-tex... t-t-text-t-t-... gimme a damn bitmap!"

Re:Good thing this is a workstation card

[toddestan]

Actually, I'm surprised that we don't see more graphics cards with larger amounts of video ram. Memory is cheap, and looking at budget cards online, usually the 256MB version of a card is less than $10 more than the 128MB version. Sure, the extra memory won't make a big difference, but it won't hurt, and most people will spend the couple of extra bucks "just in case they need it".

With the high end, I'm sure 1GB of fast memory isn't cheap, but compared the cost of high end video cards ($400-$500+), it won't increase the price that much. And I'm sure every 1337 g4m3r would drool over a gamer card with 1GB of memory (even if it showed no real improvement over 256MB), which is why I'm surprised that neither ATI or nVidia has decided to cater to this audience.

On a simular note, I'm also surprised that harddrive caches are still so small. Cache memory on a harddrive doesn't even need to be that fast either. I'm surprised that some manufacturer hasn't bothered to add $5 worth of memory to one of their high end drives, and start another numbers game going.

Re:Good thing this is a workstation card

[billcopc]

It would make business sense to make a 4gb vram graphics card because some bag of hammers would happily blow a couple grand on that slab of pointlessness. Only a handful, and I do mean 0.01% of the high-end customer base would really find a proper use for all that vram. Maybe if someone's doing very high-res renders, they might be able to justify the huge texture maps. For real-time graphics there is little sense in using gigabytes of graphics to render a little over a million pixels per frame. More efficient rendering techniques and procedural texturing would be immensely more useful IMHO.

BIOS AND GPU

[Anonymous Coward]

How about NVIDIA OS?

In other news...

[towsonu2003]

... Ati informed that, using proprietary drivers for this hardware, Linux users will be stuck at 64MB Memory with 2D support only.

\\:D/

This is anthrocentric

[Anonymous Coward]

I'm tired of hearing this anthrocentric nonsense about chips.

GPUs are not faster than CPUs because the engineers can "concentrate on one area" instead of "spreading their work around". It's not that the floating point performance of the x86 would be faster if only Intel had the time to pay attention to it. That's ridiculous.

GPU tasks are highly parallel. CPU tasks are not. nVidia can toss 24 pipelines onto a chip and realize a huge performance gain. Intel can't, because much of the time those pipelines will be empty waiting for the results of the other lines.

This fundamental difference is what separates the two domains, not it being "easier to build something that performs well in one area, than to build something that does everything amazingly well (without costing the earth to buy it)."

You need to keep your science and your homey folk wisdom separate.

Re:This is anthrocentric

[somersault]

But broadly speaking, science is still about tradeoffs etc. If Intel or AMD started developing GPUs to integrate into their chips, they would have to spread their resources thinner, unless they just bought in more people (but that's still trading one resource for another). It *is* easier to design a system that only does graphics, or sound, or networking, or any single task, rather than a generic processor. Yes, graphics calculations can be processed in parallel much more effectively than the event driven application logic that most apps tend to use, but for most non-trivial situations/problems in life, you can either just make a general product that does general things at an acceptable rate, or you can make a product that does one thing, and does it well. There is some overlap in there, depending on how you spread your resources (and how much overall time/money you have to throw at everything). Homey folk wisdom tends to work for everything..

Re:This is anthrocentric

[jcarter]

"This is anthrocentric
by Anonymous Coward"

..I'll say.

Re:This is anthrocentric

[poolmeister]

homey folk wisdom..

brilliant!

Re:This is anthrocentric

[Anonymous Coward]

Homey folk wisdom tends to work for everything.. ...except for, you know, engineers, where we tend to rely more on fundamental principles of physics and mathematics. Besides, if you think homey folk wisdom works in computers, look at what happened to Gateway. Didn't they have cows in their ads?

Re:This is anthrocentric

[Hal_Porter]

Well I asked my Folk Wisdom Consultant and he said

"Well, the cow, is not a smart animal. The ol' he-coon walks just before the light of day."

I believe this resolves your issue with the use of homey folk wisdom in engineering.

Re:This is anthrocentric

[Anonymous Coward]

Intel can't, because much of the time those pipelines will be empty waiting for the results of the other lines.

Technically, all pipelines should be doing something. In most cases they're performing speculative execution, which may or may not be needed. As you add more pipelines, you're guessing futher into the future. This means there's a higher chance you'll never use the results, so the payoff drops dramatically. The net result is that a majority of the pipelines might as well have been empty, but it's not fair to say they're just waiting. Of course, the end result is that with code execution, adding pipelines doesn't increase processor performance, which was your main point.

Re:This is anthrocentric

[somersault]

That says it so well *nod* The industry is trying to move under cover of darkness, and tries to play dumb in the light of the public eye.. yes.. >.>

Re:This is anthrocentric

[somersault]

I've seen engineers here saying it's all about compromises, and it's true. Hardness/brittleness, strength vs weight (some materials are light and strong yes, but to make them stronger you can only really add more material, or heat treat them to make stronger, but yet they will be more brittle - all compromises and trade-offs), acceleration vs speed (gearing). Most things work to the laws of that particular piece of 'homey folk wisdom'. Engineers are allowed to be wise too o_0

Speaking from a purely

[KingBahamut]

Support standpoint, we at the ubuntuforums find the support of ATI cards to be very frustrating. Their drivers dont work , and when they do, its spotchy at best.

http://ubuntuforums.org/showthread.php?t=148531
http://ubuntuforums.org/showthread.php?t=122094
http://ubuntuforums.org/showthread.php?t=148415
http://ubuntuforums.org/showthread.php?t=141090
http://ubuntuforums.org/showthread.php?t=137343
http://ubuntuforums.org/showthread.php?t=76147
http://ubuntuforums.org/showthread.php?t=75001

This is probably the largest complaint we get on the Ubuntu Forums and the UDSF(http://doc.gwos.org/ in the way of graphics cards. I think I even remember being told at one point that ATI is so driven on DirectX development that they likely dont care much about developing Open Source Drivers, or even a decent working Proprietary driver.

There have been a few Petitions to do so
http://www.petitiononline.com/atipet/petition.html
http://www.petitiononline.com/ati3/petition.html

And countless others. The community asks, almost begs, and all ATI does it laugh.

Its sad, really sad.

http://wiki.cchtml.com/index.php/Main_Page

Cell?

[Bombula]

I'm no computer expert - I'm a minority slashdot reader! - but isn't one of the selling points of the Cell processor supposed to be that it is modular in a simple way, so you can add 'cells' and gain parallel processing by offloading onto co-processors with much more ease than the traditional CPU + xPUs architectures?

Sorry if this is totally off or really dumb.

Re:Cell?

[Anonymous Coward]

And this is why the Cell is not everything it's cracked up to be and looks to be 99% hype.

Re:Cell?

[Hal_Porter]

The problem with the CELL is that the the SPE's - the small processing units that you can add easily have a lot of issues. They don't have access to the host CPU's memory. They do have a small amount of local memory, 256K IIRC though. They are also a lot simpler than modern CPU's - they execute instructions in order for example. The problem is that it's hard to take a game designed for a traditional CPU, which runs a single thread very well, and has access to lots of memory, and partition it into enough pieces to keep all the SPE's busy.

I'd think something like Sun's Niagara might be a good idea, but only for servers. For games, I think even an XBox 360 might be too far along the 'have more but simpler processors' axis.

Not render farms

[ayeco]

"meant for things like CAD or Render farms" - I know what you mean, but to be picky, these are not for render farms. They are for accurate and precise real-time rendering, the task of modeling and visualization, not the on the back-burner (no pun intended) rendering - which is what render farms do.

Speaking of backburner, the secret image below says "discreet"!!! coincidence?

Old Rule, New Rule

[Anonymous Coward]

Old Rule: Editors shouldn't be larger than the OS.
New Rule: Graphics Adapters shouldn't have more memory and suck more power than the computer.

Interesting

[Mad Ogre]

I was reading some hardware reviews on some ATI Crossfire graphics cards... the things seemed to be pretty dang good save for the fact that they couldn't go above 60hz refresh rate. So this card, they throw in a lot of memory... That's fine. But can they go above 60hz?

Re:Interesting

[Anonymous Coward]

In theory, they should be able to do way better than 60Hz. 60Hz on dual-monitor 1280x1024 at 32bit color-depth implies that 60 * 2 * 1280 * 1024 * 32 bits are being refreshed every second, which means 600MB/s throughput and with 128bit memory buses (I believe.. they might be 256bit or 64bit), that means that their memory is operating at least 40Mhz, which of course it is. In fact, at 350Mhz, the same dual-monitor 1280x1024 32bpp configuration should be able to refresh at 560Hz. Shame. Oh well.

Please don't joke about this.

[mosel-saar-ruwer]

Yes, they are 128-bit floats. They're needed for doing HDR.

PLEASE don't joke about this.

Do you have any idea how many math/physics/chem/engineering geeks would just kill for 128-bits in hardware?

It would be very, very cruel to get their hopes up like that, only to find out that you were being sarcastic...

Re:Please don't joke about this.

[Rufus211]

Sorry, but it's almost certinaly talking about 4x32bit SP floats. It's simply marketing multiplying out numbers for a bigger one.

Kids these days...

[cr0sh]

That's a lot, my first pc had16mb video ram.

My first computer (TRS-80 Color Computer 2) had a whole 16K of RAM, 8K of which was "video memory". When I upgraded to 32K, then 64K - I thought I was in "heaven". Until I got my Color Computer 3 - which had a whopping 128K of memory! WOW! Then I eventually got the 512K Disto upgrade.

The machines we (and kids) have today would have been considered so far outside the possibility of ownership back then, that they would have been mere fantasy. I dreamt about being able to program a 3D adventure game as I played Zork or even King's Quest. I even dabbled with 3D programming, but was unable to realize even the barest of my imaginations.

Today, kids have the power to realize entire worlds in marvelous and stunning 3D, replete with realistic textures and sounds. Yet it feels like they don't care to really attempt to do anything with this power, but rather consume what others give them - they would rather play in a pre-built world, than build their own. I am sure there are a few who do create their own worlds - maybe this passion will always be in the minority. I just wish it wasn't this way.

Finally, I must say I am grateful that my parents supported my computing habit at a time when home computing was still a minor thing, but a major purchase. Had they not, I wouldn't be where I am today (eh, posting to /. at work, I suppose?)...

Re:Kids these days...

[Ginnungagap42]

Tell me about it! My first was a Sinclair ZX-81 with an 8K ROM and a whopping 1K of memory (yes, that's 1024 bytes). I moved up to a Commodore VIC 20 and eventually to a Commodore 64, which lasted me throughout college.

Video games used to look like this.

when is ATI going to get OpenGL 2.0?

[heroine]

But when is ATI going to finish implementing OpenGL 2.0 for Linux? In an interview, their VP's praised their policy of starting over frequently on new code bases, but they're behind NVidia on driver support.

Sweeet!

[kannibul]

Sell off half my body organs, put that bad boy in my PC, and I can play Elderscrolls: Oblivion with all the options cranked!

Now - where can I find that Black Market...

your FIRST?!

[raygundan]

What are you, 9? Sweet jesus, man, I'm in my 20s and had a machine with less than 1MB of video memory as recently as my college years.

My *first* computer had 64k of total memory, of which only 16k could be used for video.

Re:your FIRST?!

[DavidHOzAu]

Sounds like an Apple ][.

Mine Apple ][+ had 12k ROM, and 48k RAM. (Other 4k: C100-CFFF for cards, C000-C0FF for keyboard and flip flops.) Sound familiar?

Used to have loads of fun playing Broderbund LodeRunner on it, 1 bit interlaced graphics and all. I've still got several copies of the 5 1/2" disk it came on. It'd still work on a modern emulator if I had a way of reading them.

Re:your FIRST?!

[raygundan]

It was a commodore 64. If my recollection of the mapping of the video memory is incorrect, I apologize-- I'm not quite as young as the original poster. My memory must be going.

Cell is faster

[NetRAVEN5000]

(then again, isn't that the idea of the Cell processor?)

Yes! Finally, SOMEONE gets why the Cell is faster!

Actually, the biggest thing about the Cell is that (I think) it uses vector processing, which is the same technique that your GPU uses. Vector processing is quite a bit faster than scalar processing (what most CPUs use) due to the fact that the vector processor can run operations on multiple data elements all at the same time.

And that's why I'd recommend that anyone who doesn't care that much about Halo 3 wait for the PS3 - it'll likely have noticeably better graphics than the 360 due to the fact that it can do a lot more math faster. That means it'll be able to not only have better graphics, but also to do all kinds of nice physics processing with the cycles it saves.

What this card is for -

[archiDORK]

This card is for Design Content Creation (DCC) and very large Mechanical model assemblies.

The last textures DVD I bought had textures on average around 50Mb per map (png - ca 5400p square) which then should be multiplied by 2.5 -3 for bump maps and other modifier maps. That brings us to around 125Mb to 150Mb per mapped surface. A typical scene might have 12 to 20 textures (not all of that density). This brings us to 1Gb quite quickly without the models geometry which is likely to be more than 50Mb. The128 floating point gives more accurate and less muddy colors/edges/depth when blending bitmaps.

I have been reducing these maps to 1200p 2400p wide in Photoshop to run them on my hardware. (X2 4200, 3Gb RAM, 256Mb FireGL.) I can see a time in the near future where my clients will be quite pleased with even juicier images that high resolution texturing permits. This card will no doubt pay for itself in just a few jobs, in both faster renders and higher quality product.

Mechanical design often has model sets that easily range into the 100s of Megabytes. This card will allow one to work on more complex assemblies and models. This is not the field that is at the center of my interest but even the simple I have worked on in Rhino easily hit 100Mb. Details dont come for free.

Lastly the Quattro and FireGL lines provide what gaming cards never will that is accurate display of demanding workloads. Consumer level cards provide fast display of DirectX graphics and can work well for some content creation applications (such as Photoshop), but the custom drivers (read increased speed / increased stability) and increased Floating point accuracy that workstation cards provide make them indispensable for those that make the content.

bah!

[thegnu]

The high clock rates of these new graphics cards, combined with full 128-bit precision and extremely high levels of parallel processing, result in floating point processing power that exceeds a 3GHz Pentium processor by a staggering seven times

Yeah, but with integrated graphics, you take a real performance hit.

640k

[Anonymous Coward]

640k is enough for everyone.

  / Billy Boy

nVidia SLI = 1 gig

[Anonymous Coward]

This isn't news. You have been able to build a 1 gig rig with 2x nvidia gayforce for some time now. ati firegl isn't a gaming card anyway, it's a slow perfomer no matter the amount of memory.

typo

[emarkp]

one for each primary axis

oops.

well, no...

[YesIAmAScript]

What a strange definition.

These are processes that need to be done in order to complete another process. It doesn't mean they are parallelizable. If I build a house, I need to put up 4 walls before I can put the roof on and finish it. Does that mean building walls is a parallel operation? Not necessarily, there's no reason I can't build all 4 sequentially.

Some of the setup is the same for all pixels. The polys have to be transformed to the view position space. This is done by transform and lighthing hardware nowadays and is VERY parallelizable.

Then you have to start rendering all the polys. But it is very efficient to render adjacent pixels (and adjacent polys if using a triangle strip) at the same time. That means that it isn't efficient to render adjacent pixels in independent parallel operations.

You can do some operations in parallel, many others you cannot. And the argument that you can do pixels in parallel efficently and trivially is just way off. Primarily the argument that modern rendering is similar to ray-tracing in parallelization is way way off. Ray-tracing is massively parallelizable, you can farm out pixels over even a low-speed network in an almost trivial fashion. Scanline rendering is not the same way.

Look up papers on PixelPlanes (setting the wayback machine to 1989) to see the challenges of scanline rendering in parallel.

While we're ot on this

[drachenstern]

I'm driving an '05 Chevy Malibu that has enough get up and go for me to be concerned with (two speeding tickets in less than 18 months, and didn't realize I had pushed up that high, both times were for passing on the freeway too :( ) and I get 32 city and 36 hwy, on average. If I gun a lot, I still get about 27 city and 30 hwy.

Just thought I would be a little helpful for those who are looking.

My problem is I WANT the Civic hybrid? Just can't afford it yet, ya know, with a backorder etc.

Short sighted

[TACNailed]

Who's the fool that tagged this 'overkill'?