leo@sunybcs.uucp (Leo Wilson) (08/01/87)
In article <8708010521.AA25524@ucbvax.Berkeley.EDU> FXDDR@ALASKA.BITNET writes: >Recent talk about a 16 MB ram disk and how long it would last between random >bit hits made me wonder about the memory reliability issue. >several years) in each word. I recall a business pundit in the early days of >the IBM PC who claimed that the PC was the only machine that could be used >in business because it had parity-checked RAM. After all, you wouldn't want >a bit to change in your ledger where it could cause a 2**n dollar error. If my own memory serves me correctly, IBM PC's don't actually use the parity bit, it's just there. Seems I remember some people simply leaving the chips out, or moving their flaky/bad chips to the parity bit socket because it didn't matter at all. Did any of the software use the 'extra' bit? I may be mistaken, it's been quite a while since I touched a PC, and I was more into learning about VAXen and UNIX at the time... ----- Leo E. Wilson 364 West Delavan Avenue Buffalo, NY 14213 (716)883-7573(leo@gort.cs.Buffalo.EDU)...!sunybcs[!npdp1[!leow]]!leo
kleef@cs.vu.nl (Patrick van Kleef) (08/04/87)
In article <4401@sunybcs.UUCP> leo@gort.UUCP (Leo Wilson) writes: >In article <8708010521.AA25524@ucbvax.Berkeley.EDU> FXDDR@ALASKA.BITNET writes: >>Recent talk about a 16 MB ram disk and how long it would last between random >>bit hits made me wonder about the memory reliability issue. >>several years) in each word. I recall a business pundit in the early days of >>the IBM PC who claimed that the PC was the only machine that could be used >>in business because it had parity-checked RAM. After all, you wouldn't want >>a bit to change in your ledger where it could cause a 2**n dollar error. > >If my own memory serves me correctly, IBM PC's don't actually use the parity >bit, it's just there. Guess you never suffered any 'parity errors' on an IBM-PC like I have.. Great thing, that parity check. Simply hangs up the machine once a seamingly flaky ramchip has been located. Sure, it prevents multi-billion dollar errors in your spreadsheet from occurring... it just stops the program (the whole machine, in fact) and therefore kills any data you've entered. Lots of ppl in the PC world therefore use a program called parity.com or nocheck.com in their autoexec.bat file to eliminate that check. Paul . . .
alastair@geovision.UUCP (Alastair Mayer) (08/10/87)
In article <4401@sunybcs.UUCP> leo@gort.UUCP (Leo Wilson) writes: >If my own memory serves me correctly, IBM PC's don't actually use the parity >bit, it's just there. Seems I remember some people simply leaving the chips >out, or moving their flaky/bad chips to the parity bit socket because it >didn't matter at all. Did any of the software use the 'extra' bit? I may I'm afraid your memory doesn't serve you correctly. The parity bit most definitely *does* matter in IBM PC's and at least some of the clones, if not all. There is hardware in the memory circuitry to check the parity and generate an interrupt in the case of an error. (Been a while since I soldered a PClone together, I don't remember the chip #s). Some of the clones have a DIP switch or jumper setting that lets you disable this if you don't want to install the extra chips. One of the irritating things about the IBM boot up sequence was that, while it did do a memory check, it didn't tell you *where* the problem was if it found one, it just stopped. Some of the clones have friendlier startup checks. -- Alastair JW Mayer BIX: al UUCP: ...!utzoo!dciem!nrcaer!cognos!geovision!alastair "He sure looks like plant food to me." -- Audrey II