home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!cs.utexas.edu!torn!nott!cunews!revcan!sidus!atronx.OCUnix.On.Ca!delfax!angband!Vic
- From: Vic@angband.delfax.OCunix.On.Ca (Vic Lewington)
- Newsgroups: comp.sys.amiga.hardware
- Subject: Re: ECC Was: Re: Moving 500 to...
- Distribution: world
- Message-ID: <Vic.01q5@angband.delfax.OCunix.On.Ca>
- References: <H0mXr*h41@dynam.adsp.sub.org> <1992Nov11.001328.10434@elroy.jpl.nasa.gov>
- Date: 16 Nov 92 20:59:35 AST
- Organization: -Chaos- (East)
- Lines: 25
-
- In article <1992Nov11.001328.10434@elroy.jpl.nasa.gov> ceg@bret.jpl.nasa.gov (Chuck Goodhart) writes:
- >By my calculations, 12 bits are required for 1 bit correct, 2 bit detect
- >on 8 data bits. 10 bits will only get you 6 data bits.
-
- Just for info, the systems I take care of at work use error correction in the
- processor RAM and in the RAM disks; both provide for single bit error correction
- on the fly and detection of double bit errors (which promptly bring things to a
- screaming halt.)
- The proc's RAM is 16 bits wide, with one parity bit and 5 hamming bits for ECC,
- for a total of 22 (essentially 27% of the RAM is used for error handling).
- The disk's RAM is 32 bits wide, with 7 ECC bits for a total of 39 (about 22%
- used, not much better)...
-
- Considering the cost of 1M x 4 ZIPs, I'll take my chances... ;)
- And if a normally stable app/system starts to bomb inexplicably, well...
-
- >Chuck Goodhart, ceg@bret.jpl.nasa.gov
-
- --
- Vic Lewington: "I pull out the plank and say
- (Vic@angband.delfax.ocunix.on.ca) thank you for yanking me back
- to the fact that there's always
- See, there's this HUGE parrot, and... something to distract..."
- - Kate Bush
- --
-