nagel@paris.ics.uci.edu (Mark Nagel) (11/08/88)
In article <4301@phoenix.Princeton.EDU>, zimerman@phoenix (Jacob Ben-david Zimmerman) writes: | |...okay, ignorant question here...I read earlier on that during writes, |the optidrive does TWO passes, one to zero the fields, and one to write |in the 'one' state bits. Now, since it has to do two passes and needs |to therefore know how big the files is gonna be, can it handle real-time |contintuous 16-bit input? Or will its having to go back twice choke it |up? Could a rountine be written that used the main RAM as a buffer |while it was doing its double write? Would it be fast enough to get |everything written to disc AND 'catch up' before the RAM was consumed? I believe that the concept of "file" is unknown at the level you are concerned with. The disk will do passes for each *block* it writes to disk. In essence, this is two passes per file, but happens much differently. From a user level, this is no different than any other disk, although perhaps a bit slower, since each block requires two passes to be written. Mark D. Nagel UC Irvine - Dept of Info and Comp Sci | The probability of someone nagel@ics.uci.edu (ARPA) | watching you is proportional to {sdcsvax|ucbvax}!ucivax!nagel (UUCP) | the stupidity of your action.
jdi@camelot..Berkeley.EDU (John Irwin) (11/20/88)
In article <900@paris.ics.uci.edu> nagel@paris.ics.uci.edu (Mark Nagel) writes: +In article <4301@phoenix.Princeton.EDU>, zimerman@phoenix (Jacob Ben-david Zimmerman) writes: ++ ++...okay, ignorant question here...I read earlier on that during writes, ++the optidrive does TWO passes, one to zero the fields, and one to write ++in the 'one' state bits. Now, since it has to do two passes and needs ++... + +... +are concerned with. The disk will do passes for each *block* it +writes to disk. +... Actually I heard from a very reliable source that it takes three passes to write a block. Thus reading is three times as fast as writing. The source also quoted the read speed as being something close to 900kB/s. (using the new nomenclature :-) -- John
woan@cory.Berkeley.EDU (Ronald S. Woan) (11/21/88)
>Actually I heard from a very reliable source that it takes three passes to >write a block. Thus reading is three times as fast as writing. The source >also quoted the read speed as being something close to 900kB/s. (using the >new nomenclature :-) Yep, it takes one to erase, one to write, and one to verify; however, the erase is done when the machine isn't doing anything else, so it only takes two when the time comes to write, usually. Ron
norm@cfctech.UUCP (Norm Meluch) (11/23/88)
In article <7609@pasteur.Berkeley.EDU> woan@cory.Berkeley.EDU.UUCP (Ronald S. Woan) writes: >>Actually I heard from a very reliable source that it takes three passes to >>write a block. > >Yep, it takes one to erase, one to write, and one to verify; however, the erase >is done when the machine isn't doing anything else.... Does this mean that a "DET" (disk erasure table) is kept somewhere on the disk to donote where the computer has had the spare time to erase, or is a FOURTH pass necessary to check that the area that the computer is about to write on is clear?