OK, I just briefly checked out Tom's recent article comparing 140 chipsets or something like that and noticed an interesting "observation" as follows:
"Here are a few examples: the VIA KT133A chipset accesses a PCI 33 transfer mode that achieves 33 MB/s.....Here, we'd like to address a question asked by readers who use a RAID controller (Promise or HighPoint) in a setup consisting of an outmoded chipset that has much too low a transfer rate between the Northbridge and Southbridge. In this case, the internal data transfer of the chipset is at its limit, so there's no noticeable performance increase when you create a RAID configuration. In addition, the other components such as USB and sound reduce the bandwidth, as well."
So, I have a KT133A northbridge connected to a 686B southbridge. These are connected by the "33 MB/s" PCI bus? According to Tom, all my PCI transfers should then be limited by this bottleneck? I have a two disk RAID0 array, so should this be limited to 33 MB/s??
Why I ask is that the array benchmarks much faster than this. ~70MB/s xfer rate with Winbech and Sandra says the array gets an average sequential xfer rate of ~55 MB/s. So of course these observations kind of conflict.
I would also ask too, what the heck is the point of ATA66/100 controllers on chipsets limited by the 33 MB/s xfer rate cap? In all honesty, I have never paid much attention to PCI bus xfer speeds, as I assumed my current setup didn't need more bandwidth. Thus, I am thinking the article at Tom's had some incorrect information.
"Here are a few examples: the VIA KT133A chipset accesses a PCI 33 transfer mode that achieves 33 MB/s.....Here, we'd like to address a question asked by readers who use a RAID controller (Promise or HighPoint) in a setup consisting of an outmoded chipset that has much too low a transfer rate between the Northbridge and Southbridge. In this case, the internal data transfer of the chipset is at its limit, so there's no noticeable performance increase when you create a RAID configuration. In addition, the other components such as USB and sound reduce the bandwidth, as well."
So, I have a KT133A northbridge connected to a 686B southbridge. These are connected by the "33 MB/s" PCI bus? According to Tom, all my PCI transfers should then be limited by this bottleneck? I have a two disk RAID0 array, so should this be limited to 33 MB/s??
Why I ask is that the array benchmarks much faster than this. ~70MB/s xfer rate with Winbech and Sandra says the array gets an average sequential xfer rate of ~55 MB/s. So of course these observations kind of conflict.
I would also ask too, what the heck is the point of ATA66/100 controllers on chipsets limited by the 33 MB/s xfer rate cap? In all honesty, I have never paid much attention to PCI bus xfer speeds, as I assumed my current setup didn't need more bandwidth. Thus, I am thinking the article at Tom's had some incorrect information.
