Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 16 Sep 1995 13:11:16 -0700 (PDT)
From:      "Rodney W. Grimes" <rgrimes@GndRsh.aac.dev.com>
To:        pst@shockwave.com (Paul Traina)
Cc:        hackers@freebsd.org, current@freebsd.org
Subject:   Re: looking for REALLY good hardware diagnostics
Message-ID:  <199509162011.NAA01803@GndRsh.aac.dev.com>
In-Reply-To: <199509161825.LAA07664@precipice.shockwave.com> from "Paul Traina" at Sep 16, 95 11:25:30 am

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> I've got two systems, one an early-model pentium, the other a Cx486DLC both
> experiencing the occasional odd failure when running under FreeBSD and I want
> to double-check the hardware on these machines.

Given that I do this day in day out 7 days a week I can lend some hand
in diagnosing hardware related failures, I don't go out of my way to do
this for others any longer as I have enought of it to deal with here,
but I see you making some assumptions that from my data are incorrect.

> (Yes, I know about the cache weirdness on the 486DLC, I've even disabled the
>  internal cache completely as part of my testing).

Good, that eliminates that screwwy mess :-).

> I think the Pentium either has a bad CPU (likely)
                                           ^^^^^^^^
very unlikely, in the last 2 years I have seen 0 (yes, 0) defects in a
Pentium CPU chip, if the thing will power up and load a kernel it is
good from my data.

> or a bad cache chip (unlikely)

Very likely, in the last 120 days I have had to replace 3 SRAM cache
chips that would causes system crashes, sig 11's or other strangeness.
Infact bad ``cache'' is my #1 failure mode of incoming motherboard
products.  And my #1 failure during burn in as far as electronic components
go (overall #1 is disk drives that die during my 720 minute hour glass
seek/random seek pre-burn in test :-().

> and the DLC either has a bad cache chip (likely) or bad dram (unlikely).

Dram is my #2 electronic failure, most often occurs during initial incoming
inspection during a 3 hour make world pass using a 100Mhz PB Pentium that
can really knock snot on the memory subsystem.

> Does anyone have ANY pointers whatsoever to a really really really good and
> thorough set of diagnostics that could be used to check for hardware faults?

There are non other than a VLSI tester for memory chips and simms, and ICE
for MB problems, so unless you have access to a multi million dollar test
equipment lab it is real tough.  Is what I use here is ``make world'',
if it fails that I use my years of correlating FreeBSD panics or signals
to mostlikely hardware component, and a deep gut feeling for what I have
seen over the years.

> Specificly, anything that can be used to diagnose external caches, memory,
> (and in the case of the pentium, perform cpu diagnostics) would be cool.

If it runs ``make world'' it is good, I have found no better test of
fucntionality than this as far as MB/CPU/Cache/Memory goes.  If it fails
I use the big shot gun approach and start replacing each one of those
with ``known good'' units (another benifit I have that you may not is I
have ``known good'' units around.)

-- 
Rod Grimes                                      rgrimes@gndrsh.aac.dev.com
Accurate Automation Company                 Reliable computers for FreeBSD



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199509162011.NAA01803>