Sapphire Hybrid Radeon X800 XL
The Sapphire Radeon X800 XL
To give you a clear picture of where the Radeon X800 XL falls in the 3D graphics food chain, we've put together a simple chart outlining the main features and theoretical peak fillrates / memory bandwidth of most of today's popular GPUs.
As you can see, the Radeon X800 XL falls somewhere in between the 16-pipeline Radeon X800 XT and 12-pipeline Radeon X800 Pro. The Radeon X800 XL's 16-pipeline GPU, which is clocked at 400MHz, give the card a higher peak fillrate than the Radeon X800 Pro (6.4GPixels/s vs. 5.7GPixels/s), and because the X800 XL is also equipped with higher clocked memory than the X800 Pro, the XL's peak memory bandwidth is on par with the standard Radeon X800 XT at 32GB/s (actual clock speed was 980MHz). When compared to NVIDIA's high-end offerings, the Radeon X800 XL also fares pretty well. The X800 XL has the same theoretical peak fillrate as a GeForce 6800 Ultra, and the same amount of total memory bandwidth as a GeForce 6800 GT.
Sapphire's Hybrid Radeon X800 XL looks somewhat different then the ATI built X800 XL we reviewed back in December, although most of the differences are only skin deep. Sapphire's card is built upon a blue PCB, but other than the color it is identical to ATI's reference design (Note: some early Sapphire X800 XLs are built upon a red PCB). The Sapphire Hybrid Radeon X800 XL also uses a different cooler than the ATI built card to cool its 400MHz X800 XL GPU. Sapphire's cooler is much smaller than ATI's, but it is built entirely of copper and is equipped with a high-speed 13-blade fan. We found this cooler to be very quiet during testing, and it seemed to do its job very well considering we were able to successfully overclock this card well above its stock speeds. The 256MB of 2ns Samsung GDDR3 RAM is distributed evenly on each side of the card, 4 chips on the front and 4 on the back, but only the RAM on the reverse is equipped with heatsinks. This isn't a problem, however, as GDDR3 memory doesn't get terribly warm during normal operation.