This message is from the T13 list server.
The ATA CRC is fine for the original ATA protocol, which limited command transfer lengths to 256 sectors (128 KBytes). The problem is that the new extended addressing allows much longer command sizes (e.g. 32 Mbytes as an example). For that I'd suggest forcing periodic breaks (say every 128 KBytes). While there is some overhead in getting into and out of a DMA burst, it is pretty small - done once every 64 Kbytes or 128 Kbytes and you will not notice it. Jim -----Original Message----- From: Pat LaVarre [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 28, 2001 1:08 PM To: [EMAIL PROTECTED] Subject: RE: [t13] Command completion of an UDMA OUT burst ... This message is from the T13 list server. With one burst per command, that means one Crc per command, yes? Doesn't the Crc then grow weaker as the sectors/command increases? Anybody got an idea how much weaker how fast? (((I hadn't before now heard that lots of hosts choke over a Crc per block. I had been told that an Atapi Pio INTRQ per block offers increased immunity to ground bounce, crosstalk, etc.: the extra handshakes make a system observably more robust when you're assaulting it with electrostatic charges.))) Pat LaVarre >>> Matt Rooke <[EMAIL PROTECTED]> 11/28/01 02:00PM >>> ... > In addition, I > think that most HDDs won't initiate burst termination before all data for a > command is transferred. True statement for all modern HDDs. Fujitsus first generation UDMA chip terminated after each sector, and this did break some controllers (who didn't read the spec). If there were still drives doing this, we would likely hear about it. ... Subscribe/Unsubscribe instructions can be found at www.t13.org. Subscribe/Unsubscribe instructions can be found at www.t13.org.
