Sapphire X800 XL Review

Delivered Potential?

by Josh Walrath

The X800 XL Technology

            In December ATI released the specifications on a whole gaggle of products, which were the eventual refresh of the X800 Pro and X800 XT series of products.  The new chip that the X800 XL is based on is the R430, which is essentially a R420 (that which powered the X800 Pro and XT series), but built on TSMC’s 110 nm process.  The 110 nm process is a size optimized process, and does not appear to be as fast as TSMC’s 130 nm Low-K process.  Apparently yields and speed bins for this process are excellent, which is why we are seeing the $199 to $299 price points for products based on this chip.  Other than the switch in the process used, the chip is essentially the same as the R420.  No major features have been added or changed.  This means that Pixel Shader 2.0b and Vertex Shader 2.0 are supported by the hardware.

While this picture turned out to be a bit more blurry than I wanted, the size difference between these two chips is still quite obvious.  The top is the R420 (X800 Pro) and the bottom is the R430 (X800 XL).

            The X800 XL is a full 4 quad pixel pipeline design (16 pipelines) with 6 vertex shaders running at 400 MHz core.  The memory is on a 256 bit bus and is specified to utilize 256 MB of memory running at 490 MHz DDR (980 MHz effective).  The overall specifications indicate that it should be a GeForce 6800 GT killer, but in testing it appears as though the NV40 architecture is a bit faster per clock than the corresponding ATI products.  From a fillrate point of view, the X800 XL does beat out the 6800 GT, as well as its X800 Pro counterpart.  The raw fillrate of the X800 XL at spec is 6400 MP/S, while the 6800 GT is 5600 MP/S, and the X800 Pro is 5700 MP/S.  This should give the X800 XL a significant advantage over both of its counterparts, but that is not always the case.  The 6800 GT is just slightly faster than the X800 XL in many applications, but the X800 Pro is simply outclassed by the X800 XL (as will be shown in this review).  This confuses the situation a bit for consumers, as the X800 Pro typically retails for much more than the X800 XL.

Just for grins I dug out ATI's first 256 MB card, the Radeon 9800 Pro 256.  ATI's PCB designers certainly do a nice job with their layout.  The X800 XL packs a lot more punch in a smaller package.

            After taking off the heatsinks of both the X800 Pro and X800 XL, I did a quick and dirty measurement of the exposed die.  Now, this should not be an official size of either of the dice, but it gives a pretty good idea of the difference in size that the 110 nm process will give.  The X800 Pro measured 16 mm x 16 mm in size, which gives a total area of 256 mm square.  The X800 XL is around 14 mm x 14 mm, which comes up as 196 mm square.  On a 200 mm wafer, this comes out to be around 96 gross die per wafer for the X800 Pro, and around 120 gross die per wafer for the X800 XL.  Each wafer costs around $1,000 US to produce, but we don’t know the actual cost for the parts as both ATI and NVIDIA deal with TSMC on a “per good die” basis, but the estimated extra 24 dice on the cheaper process does allow for lower costs per die.

Next: The Sapphire X800 XL Board

Copyright 1999-2005 PenStar Systems, LLC.