Many people have been wondering about how the Parhelia will benchmark. I'm included in that group. Once I heard that Matrox said not to expect the card to win any awards at Quake III FPS contests, I realized that Matrox was in an uphill battle to win the hearts of the reviewers who are so often stuck on such measurements. I began to wonder about what market(s) Matrox was really going after with this card. So I asked and Matrox responded.
I also threw in a bonus question about z-compression and related techniques that seem to be missing from the Parhelia.
Here are the quotes (to avoid being a whore. )
For those who are interested, the rest of the article is <a href="http://www.cypherlabs.com/modules.php?op=modload&name=News&file=article&sid= 18&mode=thread&order=0&thold=0">available</a>
I also threw in a bonus question about z-compression and related techniques that seem to be missing from the Parhelia.
Here are the quotes (to avoid being a whore. )
The target markets for Parhelia-512 board level products will be the workstation user due to our highly flexible multi-monitor implementation with DualHead-HF, focus on 2D image quality and implementation of top notch components like 5th order filters as well as stability. We are also targeting the PC enthusiast who is looking to play games across three monitors in Surround Gaming and who utilizes his or her PC for more than just gaming, like for image editing with 10-bit color and gamma correction and is looking for the overall high fidelity graphics experience that Parhelia-512 board level products can offer.
Parhelia-512 does have some z-optimizations, like fast z-clear, but due to the chips rather large memory bandwidth, because of its 256-bit DDR memory interface, we are confident that we will be able to handle z calculations efficiently.
Parhelia-512 does have some z-optimizations, like fast z-clear, but due to the chips rather large memory bandwidth, because of its 256-bit DDR memory interface, we are confident that we will be able to handle z calculations efficiently.
Comment