Hi, I just read this in www.ocworkbench.com forum and thought this will be interesting to many of you.
quote:
"I got to test the SiS735/745 in the context of my work. We're developping video editing platforms, and these babies are real PCI-bandwidth hogs. (720 x 240 pixels x 60 fields per second is a lot of data!)
We wanted to support AMD AthlonXP CPUs really hard, since these processors are amazing computing powerhorses. However, the Via chipsets clearly held the platform back, with PCI bursts scoring no higher than 40 MB/s when two bus master devices try to perform simultaneous transactions (IDE does count for one!). This could get pretty technical, but let's just say that in these particular tests, Via scored 40, Intel 90 and SiS managed to pull out an amazing 92 MB/s burt rate (out of a theoretical 128 MB/s).
Before someone asks about the Via issue, YES, this relates to the PCI latency bug Via has acknowledged recently, and NO, it cannot be fixed by a software patch like they would like you to believe. The PCI core has an hardware bug, and no driver nor bios update can fix this. Also, since the new KT333 will have the same PCI core as the KT266A, PCI bandwidth will be no better in the newer chipset.
That being said, KT266A/KT333 platforms perform just fine when equipped only with an AGP card looping Quake3 demos. That's exactly why most hardware review sites sing praises for the KT266A. But, honestly, is this really all you intend to do with your computer ? Not me "
quote:
"I got to test the SiS735/745 in the context of my work. We're developping video editing platforms, and these babies are real PCI-bandwidth hogs. (720 x 240 pixels x 60 fields per second is a lot of data!)
We wanted to support AMD AthlonXP CPUs really hard, since these processors are amazing computing powerhorses. However, the Via chipsets clearly held the platform back, with PCI bursts scoring no higher than 40 MB/s when two bus master devices try to perform simultaneous transactions (IDE does count for one!). This could get pretty technical, but let's just say that in these particular tests, Via scored 40, Intel 90 and SiS managed to pull out an amazing 92 MB/s burt rate (out of a theoretical 128 MB/s).
Before someone asks about the Via issue, YES, this relates to the PCI latency bug Via has acknowledged recently, and NO, it cannot be fixed by a software patch like they would like you to believe. The PCI core has an hardware bug, and no driver nor bios update can fix this. Also, since the new KT333 will have the same PCI core as the KT266A, PCI bandwidth will be no better in the newer chipset.
That being said, KT266A/KT333 platforms perform just fine when equipped only with an AGP card looping Quake3 demos. That's exactly why most hardware review sites sing praises for the KT266A. But, honestly, is this really all you intend to do with your computer ? Not me "
Comment