Architecture & Speed

[Xmo1]

<B>The Celeron 566, 66mhz FSB, 100mhz ram, 143mhz video with AGP4, UATA66/100, PCI, ###</B> I bought a UATA100 hard drive but my MB only supports up to UATA 66. The solution I'm told is a UATA 100 PCI card. My PCI speed looks like it runs at 66/2 or 33mhz. The claim is that it brings a UATA66 up to UATA100 speeds. The drive controller is on the MB, so I could see how a UATA66 could operate at full speed on a board with a 66MHZ FSB. It seems to me that a drive controller in a PCI slot would be limited to the speed of the slot (33mhz). Next question: My video card runs at 143 mhz and the ram at 100 and the FSB at 66. Doesn't the FSB limit everything to 66mhz through the CPU? I've been looking for a generic MB architecture diagram that would explain it all, but right now I feel like I'm getting trapped behind the hype when the discussion moves around architecture and speed issues.

[spondylolisthesis]

I think the fastest IDE HDD currently for sale is the IBM DTLA series. They run at a little better than 35MB/sec (sequential xfer). So the ATA-66 interface card/controller should be able to comfortably accomodate your ATA-100 drive.

The bandwidth or throughput of the PCI bus is a little less than 133MB/sec because it has a "width" of 32 bits. So it will not be a bottleneck in the near future. See the topic "System Bus Functions and Features" on this site.

The only argument I have seen that is able to logically support the requirement for the ATA-100 interface is that the HDD cache can achieve a higher data throughput than can the electromechanical subsytems/devices of the drive therefore the overhead of the ATA-100 interface is justified, if only for the amount of data in the cache.

In my opinion, that does not justify the additional cost.

I have one of the DTLA HDD's, and have noticed that it seems to run noticeably quieter (much quieter) and faster under Win98SE/FAT32, than it does under Win2000(Advanced Server)/NTFS.

Does anyone have an explanation for this?

[This message has been edited by spondylolisthesis (edited 11-05-2000).]