Date: Thu, 13 Apr 1995 12:52:01 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: pw@snoopy.MV.COM (Paul F. Werkowski) Cc: freebsd-hackers@FreeBSD.org Subject: Re: 940804 (vaporware ;-) reboots the system either: Message-ID: <199504131952.MAA10852@gndrsh.aac.dev.com> In-Reply-To: <199504131904.PAA02469@snoopy.mv.com> from "Paul F. Werkowski" at Apr 13, 95 03:04:00 pm
next in thread | previous in thread | raw e-mail | index | archive | help
[cc: reset to freebsd-hackers@FreeBSD.org] > >>>>> "David" == David Greenman <davidg@Root.COM> writes: > > David> test). Since you seem to have relatively new hardware, it's > David> extremely unlikely that your problem is in any way related > David> to memory sizing. If I were going to troubleshoot the > > Last year I attempted to boot FreeBSD 1.1 on new hardware > and ran into all kinds of boot problems. Failure to boot > at all or unreliable boots and/or operation. DOS ran OK > and the diagnostic programs running on DOS ran OK. I > eventually went to memory simm swapping from a known good > similar system to prove that at least one simm was bad. > The vendor finally showed up with a diagnostic (called CHECKIT) > that finally did reveal bad memory just over the 1M boundary. > Given this result, he gave me new (tested with CHECKIT) simms > and I have not had a problem since. This is the second time > I have been burned by bad memory in a new box and will always > suspect that first. Don't believe any DOS diagnostic software > either!. Though CHECKIT does a somewhat reasonable job I have seen memory that will run CHECKIT for 24 hours and still fail to run Unix or Novell server. IMHO, the only real test (I guess this comes from 3 years of being a VLSI test engineer) is to run the stuff on a memory tester. I have access to one locally, and when I have my doubts about memory I take it over there and get it tested the ``right way''. These SIMM tester boxes are a little expensive for me to own one, but they sure come in handy when dealing with memory related problems. Some of them can even measure the access times and do +/-5 or +/- 10% voltage margining so you can really be sure the memory will work reliable in just about any motherboard. Trying to test memory in a PC prevents many problems as it is not an adaquate tester due to the lack of the ability to do what are commonly known as the 5 point boundary test cases. Often memory will work fine when Vcc = 5.00V and Vil/Vih are nominal, but will fail if any of the 3 parameters are moved to worst case values. Very few motherboards actually run at ``ideal'' values, Vcc can often be down 100 to 200mV by the time it actually gets to the die inside the memory chip on the SIMM. [Remeber, there are no less than 2 connectors in this path, and the number 1 cause of failure of electronic systems is bad connections (even down to the die level :-)] -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Custom computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199504131952.MAA10852>