The URL you followed to get here is no longer in use and we have redirected you. If the URL was in your bookmarks then please
update your bookmarks. If you followed the link from another web site or search engine then please notify their webmaster.
Introduction
In the last year or so we've seen a variety of new GPUs from all four major graphics IHVs. The technology push in the 3D space has been immense, the IHVs all implementing new technology, offering more performance and features, on ever decreasing process sizes and at ever increasing clocks, with huge framebuffers, dual DVI ports as standard on some models, video processing, PCI Express, and let's not forget SLI. It's hard to keep up, wouldn't you agree? And that's without delving into the specifics of modern graphics processing architectures to see how they extract their performance and implement their features.
For me, where my day job requires me to keep all that information on the go in my head at the same time, thinking about my ideal graphics card is both incredibly easy and fantastically challenging. The easy way to graphics nirvana would be to pile all the best bits from current GPUs onto one chip, place that chip onto a well-designed PCB, use Samsung's latest 512Mibit DRAM devices, implement dual DVI or HDMI outputs and have it powered entirely by my PCI Express slot, and have the cooler be resolutely one slot and very quiet.
Just imagine it. A 512MB graphics card that implements Shader Model 3.0 and beyond, has a considered architecture that balances the twin needs of vertex and fragment unit power, for both current and future titles, with a calculated number of ROP units for pushing the pixel output to the screen, that supports decoding two HDV streams on the chip in parallel, that can output them and my 3D games to any window arrangement I choose over my two LCD monitors without performance loss. I want all that in a chip that's not the size of my arse, mounted on an all-black PCB with silver detailing, using a single-slot cooler that's also black, with a glimmer of white light protruding from betwixt the black blades of the low-speed, high-airflow fan produced by white LED light. One more red, blue or green PCB with blue LEDs and I'm liable to have a breakdown.
I want adaptive AA with up to 12 programmable AA samples, and a more intelligent algorithm that doesn't rely on fixed angles for my anisotropic filtering settings, for better image quality. I want floating point texture formats to be usable with the new multi-sample AA modes. I want 10-bit per component 2D output like the S3 DeltaChrome. I want single-cycle trilinear texture filtering in multiple adaptive texture units, per shader unit, be that vertex or fragment. I want an abundance of arithmetic power from multiple full ALU units, so I can hack my own general purpose programs on the GPU and easier hide my horrible instruction issuing mistakes when I do so, so my codes are still quick to execute.
I want flawless drivers that detect my LCDs on first install, download their drivers since the Plug and Play ones in Windows aren't great, set those up, then optimally let me configure their orientation with a single mouse click. I want control over the fan speed of the card using the driver. And I want it to be able to participate in multi-GPU rendering if I decide to buy another one. Finally, I want the back of the PCB personally signed by the lead engineer that created its core design. All that has to give me at least twice the current rendering power of SLI 6800 Ultras, in that single card.
I don't want much.
Snapping back into the real world, attempting to find my perfect graphics board among the hundreds that populate store shelves and online inventories the world over, I still want single-slot, quiet cooling from a board with a small PCB, that can challenge 6800 Ultra SLI in places with performance, that has great image quality, dual DVI, and that rides the PCI Express bus. I also want it to have video input, so I don't have to use a separate piece of hardware to play my Playstation 2 on my PC's monitors.
And I can get those things today, if I'm prepared to look beyond the usual board vendors and consider something with a very intelligent cooler design. I honestly can't abide massive coolers and their obnoxious fan designs or horrible aesthetics. I want good looking, small and quiet, and I want my performance too. I've recenty
offered the opinion that X850 XT and X850 XT PE are nonsense purchases, given the price and performance of X800 XL, and that's true for 99% of you. For the other 1%, people that would seriously consider something like a Patek Phillipe watch if they had the money, X850 XT and XT PE stay relevant. I'm writing for the 1% of 1%, then, but as they say, the
high high-end defines the rest of the brand.
So when GeCube sent along their Radeon X850 XT, I was pretty sure I'd just found the
current graphics card of my dreams, given that I can just about afford to shop in the £300 space if the card is fantastic. And for the GeCube, it's all about the UniWise.
printer friendly layout 
discuss in the forums 
email to a friend
IE7 Search
Firefox 2 Search