upgradable vid cards

[res07tg2]

after reading docs posts on motherboards i thought id get everyones thoughts on this idea

what about a vid card that you can upgrade the ram on ???? or for that matter the gpu like you can on a mother board



i don't even know if this is posable such things go way past my skill leval but since this a think tank its a thought so here it shall apear

what does everyone else think let me know

[Toaster]

In the non-x86 world (non-intel/AMD etc) video functions are handled by both the main CPU and the video interface. To do what you suggest requires "compatability" between manufacturers and this won`t likly occur in the X86 world anytime soon. This would mandate a "universal" video bus, a universal "CPU --> video instruction set" and a universal video hardware to CPU/GPU bus architecture. That means straying away from the norm. This would also mandate a "standard" in the way video hardware is accessed in all modes.
Maybe, in the years to come, a company like Nvidia or some other entity will incorporate a "ziff socket" for the GPU on a board with a given amount of memory. I seriously doubt this for any time soon. You ask that someone stray from the fold and "go it alone". Only 2 companies have done this and succeded and neither are "Windows" usable. These are Silicon Graphics and Sun Microsystems.
Take for instance the SGI graphics subsystem.
This subsystem is bleeding edge fast but very specific. One card, the "MaxImpact" has numerous "planes" to accelerate graphics at every concievable angle and method. Still the fastest thing on the desktop and its almost 10 years old. Some of the systems have individual boards to do "X" graphics, "Y" graphics, a geometry engine and a local CPU/GPU to make the thing "work".
Now, dash down to Nvidia and say "sure i`ll pay 2,000.00 for a video card.
In short, thats why things are the way they are.

[Xayd]

The cost factor that Toaster pointed at is one reason I wouldn't want that to happen. Competition has done great things for the graphics/gaming market out there. With nVidia and 3dfx fighting like cats and dogs over the last year, we've seen it possible to play a PC game smoothly that looks damn near like a movie with 300 dollars worth of hardware or less.

If there were a set standard for graphics processors, who would set that standard, and how long would it take them to muscle every other manufacturer out of the market and raise prices accordingly?

And honestly, with a set of rules and standards innovation slacks off. I have several pieces of proprietary company software looking at me right now that would get an in-house programmer from a generation ago fired, but a half-working 1-day-Visual-Studio-job-for-Win9x-only piece of software is 'good enough' in today's world. Who's to say graphics processors wouldn't fall to the same lower standard if there was a generally accepted design?

If a generally accepted 'way' was devised that would serve the purposes of all conceivable applications for years to come was devised, it might be do-able. But who sets these standards for hardware and software? Microsoft and Intel right now for the most part, and Apple has their market as well. Don't forget Gates saying way back when that no single user would ever need more than 640 kilobytes of RAM on a workstation or home PC. Is this the guy we want to devise a standard for graphics processing? Nope, ;).

Take the PC Game market for instance. Has Direct X made gaming on a PC incredible in every way, for example? Actually it's just made it easier to produce a crappy game, as evidenced by the crap available at any store that sells PC Games. Truly great games still require innovation from the developers of said game, and are few and far between.

When an idea becomes practice, and eventually turns into a generally accepted standard, it isn't always because it's a great thing. And it rarely means that great things can be attributed to its invention. More often it means that crap can be produced by lesser manufactuters/developers, simply by employing this standard rather than having to do the hard work themselves.


Xayd

[Toaster]

One thing that is somewhat in line with Xayd is the recent technology that 3D graphics has undergone. In the past 2 years alone, graphics throughput has more then doubled and the price halved.
3D graphics for the Pee-Cee is reletively new for the most part and software is still reeling with the number of ways to "do the same job".
These are a few existing standards, most of which are grossly outdated and actually prohibit performance.
One standard dates back to 1987 (even earlier) is the VESA or "Video Electronics Standards Association" which simply lays out a "method" to access higher functions of a "typical" SVGA adaptor.
Now is NOT the time for such standards as Xayed suggests as the 3D relm is just getting interesting. Maybe in the coming years this would change.
On the "memory" standpoint, expandible memory for video also dates back many years and only offered higher resolutions and deeper color depths.
Remember, it was the early 90`s when 3D cards hit the market. The first few adaptors were the "Monster3D" variant or also known as "VooDoo1".
A great deal has been done so far and the fun is just getting started.
Now, video subsystems boast 64/128/256 MB of local memory.
The fun will begin when the CPU and the graphics processor completely divide into thier own forms without relying on the other. This would be the main CPU does its own tasks and the video interface is actually a self contained computer in itself needing no help from the host CPU.

[Xayd]

And along that same line, if you look at the AGP standard for video cards, it's a hindrance right now more than anything, IMHO.

In the past year 3dfx has tried SMP to squeeze more performance from graphics processors even, and failed miserably at it.

Despite the recent cheap availability of pumped up memory in the form of DDR and incredibly high-clocked CPUs, the good ole AGP bus still runs at 66mhz, due to it's links to the 32 bit PCI 'standards' ;), and it's one of the bottlenecks that hold back the further development of high end 3D gaming/editing PCs.

Xayd

[Toaster]

To make big gains in video throughput, the AGP port has to go.
This would mandate a "true" CPU bus to video bus. In the days of yore, this was called "the local bus". While the idea was sound, the actual use was far from ideal. AGP went a step further by increasing the clock rate and more closely coupling the video bus to the CPU bus.
IMHO, they have this backward.

Lets take SGI (silicon graphics) for an example.
In thier entry level 24 bit adaptor of yore, they used similar approaches as now on the Pee-Cee. However, to gain REAL bandwidth, they redesigned the video bus to work along side the CPU. Now, the CPU and the video interfaces act as one at the speed of the CPU. Couple this with a 64bit data path and performance becomes obviously fast. Next, they dedicated several "Geometry Engines" or "local processors" if you will so that the vid card is now fully independant of the CPU. The video even has the ability to control the system bus to move its data.
This could be seen as a sort of "bus master" but actually the system is totally alien to Pee-Cee`s. Now, its as if the system has a SBC (single board computer) that does video duties. This SBC has multiple "X" and "Y" and multiple "plane" acceloration. The main CPU is almost completely out of the picture as everything is handled locally by the SBC video adaptor. So, in essence, a computer within a computer.
Nvidia is on the right track so far. Thier NV20 now boasts more transistors then the Pentium 4. Still, thier graphics chipset relies on a 32/16 bit transfer bus operating at 66mhz. Reinvent the video bus because from here on out, you are looking at 5-10% increases at best.