After getting off to a less than stellar start at launch, the battle between the GeForce 7800 GTX and Radeon X1800 XT is finally beginning to heat up. Regular readers of this site probably know the story pretty well by now, but just in case we’ll provide a quick recap.
The tale in some ways actually begins over a year ago, with the debut of shader model 3.0 in NVIDIA’s GeForce 6 lineup. After trailing ATI early-on in the days of DX9, NVIDIA gambled big and jumped from the first generation of shader model 2.0 DX9 shaders straight to shader model 3.0, which, among other things added support for longer, more complex shader programs, as well as features designed to make writing 3.0 shaders easier such as dynamic branching/looping. ATI on the other hand played it more conservative, opting instead to adopt shader model 2.0b for their Radeon X800 series of graphics products.
ATI argued all along that shader model 3.0’s requirement of 32-bit precision didn’t become feasible for them until they move to 90-nanometer. Their argument was that producing a high-end shader model 3.0 part on TSMC’s existing 130-nm manufacturing process would be too costly, the die would be massive and selling it even at high-end $400+ price points wouldn’t be practical. Basically they didn’t want to compromise on their margins. At that time, TSMC’s smaller 110-nm process was used solely in value graphics products, no one at the time had a clue the clock speeds 110-nm would ultimately prove capable of achieving with high-end parts containing 200 million transistors or more. This wasn’t proven until NVIDIA’s GeForce 7800 GTX debut in June of 2005 over one year later.
ATI’s decision to wait for 90-nm probably wouldn’t have been a bad one if they could have stuck to their initial goals for the X1800, namely those being very high clock speeds (700MHz or more has been reported) with a launch date around the early summer time frame of 2005 – right in line with NVIDIA’s GeForce 7800 GTX introduction. After all, the X1800 contains a whopping 320 million transistors (roughly 18 million more than GeForce 7800 GTX despite containing fewer pixel pipelines), a 300+ million transistor chip built on TSMC’s larger 0.13-micron process probably would have been costly for ATI to manufacture. In addition, ATI’s X800 line performed well in comparison to GeForce 6800, in fact ATI took the performance crown from NVIDIA at the beginning of this year with their Radeon X850 XT Platinum Edition, only SLI and ATI’s nagging supply issues kept the X850 line from being more successful.
On launch day for the Radeon X1800 XT, the card showed signs of promise, but it was by no means a slam-dunk product. The Radeon X1800 XT 512MB was able to outperform the GeForce 7800 GTX in Direct3D titles, but OpenGL applications like Doom 3 and IL-2 Sturmovik: Forgotten Battles favored NVIDIA’s 7800 GTX. When you factored in the board’s price and availability, the general consensus was that the GeForce 7800 GTX came away as the better value.
Availability is improving on the X1800 XT and as a result, prices on retail boards are falling quickly. In addition, newer drivers from ATI have improved the X1800’s performance considerably – the X1800 XT now outpaces the GeForce 7800 GTX 256MB in both Direct3D and OpenGL applications.
With all this in mind, lets take a look at MSI’s Radeon X1800 XT board, the RX1800 XT VT2D512E.