Re: S.M.A.R.T.

Gadi Oxman (
Mon, 15 Sep 1997 01:46:23 +0300 (IST)

On Mon, 15 Sep 1997, Gerard Roudier wrote:

> I am very glad to know that IDE disks needed such a feature to become
> In my opinion, as long as they will not be able to:
> - Really disconnect the bus
> - Allow command queuing
> - Implement parity checking
> - Allow long cable
> - Etc ..
> Any complex feature will not make them SMART in my opinion.

The basic strength of IDE is a low price/performance ratio. Support for
some of the above features is improving, though:

- ATAPI tape drives are able to completely disconnect from the
IDE bus, and overlap operations with the other IDE device on
the same channel.

An IDE tape drive which shares an IDE channel with an IDE disk
drive under Linux will only prevent the disk drive from access
to the small bandwidth which is required to burst-transfer the
data from/to memory to/from the internal tape buffer, without
requiring the disk to wait for actual tape I/O completion.

- Under Linux, IDE/ATAPI cdrom drives are able to disconnect
from the IDE bus during large seek operations.

- New versions of the ATAPI specifications allow a generic ATAPI
device to disconnect during data transfers as well -- not supported
yet under Linux, and probably not commonly implemented yet.

- The Ultra-DMA/33 protocol incorporates a CRC-16 error checking
protocol which can detect errors which result from noise on the
IDE cable.