munck@chance.uucp (Robert Munck) (11/18/89)
In <3004@ganymede.inmos.co.uk> davidb@inmos.co.uk (David Boreham) writes: > >Try looking at "Flash EPROM" devices. These are either bulk or >sector erasable (most require 12v Vpp but some from ATMEL are >5v only) and can be programmed byte-by-byte. The best have >10,000 program-erase cycles and the cheapest cost about $160/Megabyte. >Much cheaper than full-featured EEPROM. > >Some new handheld computers (for instance the new one from Scion in >England) use Flash-EPROMS for simulated disk. > >Manufacturers: Intel, Seeq, Atmel. >Future manufacturers: TI, Toshiba, NEC, SGS-Thomson Microelectronics. > >David Boreham, INMOS Limited | mail(uk): davidb@inmos.co.uk or ukc!inmos!davidb >Bristol, England | (us): uunet!inmos.com!davidb >+44 454 616616 ex 547 | Internet: davidb@inmos.com Here's an interesting problem; design a file system manager to get the maximum use from a disk simulated with Flash-EPROMS. Assume "typical" MS-DOS disk usage statistics -- files added, edited, deleted. I think you'd have to assume availability of a static RAM memory with no limits on erase cycles, maybe 1/16 the size of the EPROM memory. The size of the minimum erasable block -- "sector" -- is probably an important factor. Also, I'd like to know some real price and erase cycle limit values; from the way it's phrased, I doubt that the 10K cycle chips cost $160/MB. Cute problem; I wish I was still teaching OS design so I could hand it to a bunch of bright kids as an assignment. -- Bob <Munck@MITRE.ORG>, linus!munck.UUCP -- MS Z676, MITRE Corporation, McLean, VA 22120 -- 703/883-6688
pardo@cs.washington.edu (David Keppel) (11/20/89)
In <3004@ganymede.inmos.co.uk> davidb@inmos.co.uk (David Boreham) writes: In article <9804@saturn.ucsc.edu> munck@chance.uucp (Robert Munck) writes: [Flash EPROM devices for simulated disks; $160/Mb] How about very slow, very low power, high-density, battery-backed CMOS RAMs? They might be cheaper and wouldn't bite the 10,000 cycle limit. ;-D on ( If memory serves me correct ) Pardo -- pardo@cs.washington.edu {rutgers,cornell,ucsd,ubc-cs,tektronix}!uw-beaver!june!pardo