Re: Triton DMA

mlord (mlord@pobox.com)
Sat, 29 Nov 1997 13:29:54 -0500

Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Eloy A. Paris: "Re: smbfs and 2.1.xx"
Previous message: Bill Hawes: "Re: Write to nfs fs gives I/O Error"
Maybe in reply to: Hubert Mantel: "Triton DMA"
Next in thread: H. Peter Anvin: "Re: Triton DMA"

linux kernel account wrote:
>
> On Sat, 29 Nov 1997, Rogier Wolff wrote:
>
> > You're right that an unacceptable overhead would be incurred if
> > software would need to calculate the CRC. As to the speed of
> > calculating a CRC against that of parity, both can be implemented in
> > hardware with just a few xor gates.
>
> But IS it implimented in IDE hardware? And what kind of overhead to it
> cause bandwidth wise?

It is implemented in hardware.

And, it is only a crc16 rather than crc32 (had it wrong in earlier
post).
We just need to add detect/retry logic for CRC errors,
which are signalled to us by a new bit in the IDE ERROR register.

Just cuz it's IDE doesn't mean the designers were clueless.

Cheers!

--
mlord@pobox.com	
the Linux IDE guy

Next message: Eloy A. Paris: "Re: smbfs and 2.1.xx"
Previous message: Bill Hawes: "Re: Write to nfs fs gives I/O Error"
Maybe in reply to: Hubert Mantel: "Triton DMA"
Next in thread: H. Peter Anvin: "Re: Triton DMA"