On 7/5/10 11:55 AM, Dave McGuire wrote:
There's no standard on-disk format for SMD
drives, unfortunately.
FYI, from a while ago:
--
Newsgroups: comp.sys.prime
Subject: Re: Abandoned Prime looking for new home (UK)
Message-ID: <369pu29hl789kaf5j302ve22bstihhg2av at 4ax.com>
Date: Mon, 05 Mar 2007 23:27:55 GMT
On Mon, 05 Mar 2007 10:15:02 -0800, Al Kossow <a... at spies.com> wrote:
Jim wrote:
On Mar 4, 11:49 pm, Dennis Boone <d... at
ihatespam.msu.edu> wrote:
> You do realize the drives are Century 315's, i.e. 14" SMD drives,
> right?
I guess a determined person might be able to hook
the drives up to a
Sun or SGI with an Emulex SMD controller.
Nope. You need to use a controller that understands the sector format
written by the Prime controller. Every vendor had their own way of doing
things like sector error checking/correction.
Actually what every vendor did is to alter the actual sector layout.
What was in the sector headers, data area of the sector and the error
correcting code was very standard. While Prime put the Record ID on
the front of the data portion of the sector (32 bits), it wasn't
checked in the controller at all, it was checked by the disk driver
after the record had been transferred to memory.
99% of SMD controllers used a standard sector/track/ 16 bit CRC code
for the header on the sector, and virtually every one used the same 32
bit polynomial error correcting code in the data portion of the
sector. (And everybody used the Chinese Remainder Theorem to calculate
the correction).
For the FS4 we consider using a 51 bit code, but dropped the resultant
media would not be readable by a 4005 or IDC1, nor would 4005 or IDC1
written media be readable by us. The 51 bit code handled certain types
of multibit errors much more gracefully. The 32 bit code would
miscorrect if there were two errors, and the space between them was
more than 9 bits. That is why any track that had more than one error
on it was considered defective in the CDC spec.