Date: Fri, 5 Dec 1997 18:16:07 -0500 (EST) From: Bill Vermillion <bill@bilver.oau.org> To: freebsd-isp@freebsd.org Subject: Re: FW: Touching Base Message-ID: <199712052316.SAA27330@bilver.magicnet.net> In-Reply-To: <Pine.BSF.3.96.971205143638.16807T-100000@picnic.mat.net> from Chuck Robey at "Dec 5, 97 02:44:09 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
Recently Chuck Robey said: > > -----Original Message----- > > From: Tom Holderby > > Sent: Friday, December 05, 1997 11:57 AM > > To: George M. Ellenburg > > Subject: Re: Touching Base > > > > George, > > > > [My Comments:] [...Non Relevant Material Cut...] > > > > Let me also give you a quick update on the FreeBSD situation. When I > > run > > that "tar x" while sitting at the console I see "Memory Parity Error". > > But > > I ran several iterations of the CheckIt memory diagnostic with its most > > advanced testing options and it's not turning up any errors, so I tend > > to > > believe it's a software problem. .... > Anyhow, it's not possible to test memory using a PC. The only > way to do it is with an expensive piece of test equipment. The return > from Checkit is only good for complete failures; any other indication > isn't worth anything at all. When I've been asked about checking memory in PC based machines, I tell them the best program to check memory is Unix. :-) I saw a memory test program from a board manufacturer that passed all the memory with flying colors, but would core dump in Unix. Looking at the core dump there were large areas of text, and about 1/2 the letters were of by one character. It acted like bit 0 (this first bit) was stuck on permanently. Pulled the board, and there in the first position of all the 256K chips (this was a while ago) there was a lonely 64K chip. Trust nothing. I remember from long ago that testing memory on a PC would take about 15 minutes for each 32K, complete with walking bits, checkerboard patterns, variable timing on refresh, etc. I've seen nothing like that. All of the PC memory checkers I've seen really seem only to tell you there's a chip in the socket, and as the above shows, it can't even tell if the chip is the wrong one. -- bill@bilver.magicnet.net | bill@bilver.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199712052316.SAA27330>