PCBUILD Archives

Personal Computer Hardware discussion List

PCBUILD@LISTSERV.ICORS.ORG

Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Steve Collins <[log in to unmask]>
Reply To:
PCBUILD - Personal Computer Hardware discussion List <[log in to unmask]>
Date:
Tue, 23 Apr 2002 18:01:03 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (72 lines)
Hi Brian,

I understand your approach and it is probably quite valid.  My approach,
however, is different.  Your "scorched earth" approach (I love the term
by the way - it's a lot more descriptive than baselining a system) is
tantamount to burning down the house to cure a leaking faucet.  This is
especially true if you take a less "software-centric" approach to system
troubleshooting.  My formal education and background is in electronics
first and then software development, so my approach may be different
from those who are more comfortable with the software side of things.

Step 1 visual and basic system checks (1 hour) - After doing the usual
checks (board cleanliness (anyone but me ever find a gummy bear stuck to
the inside of a case???), CPU verification (has someone put in a
rebranded CPU????), system config., amount of free disk space, number of
folders (directories), drive and fan speeds, system temp., memory
integrity (too much memory??? - Win9x!), visual inspections for
overheating components (use a magnifier and VOM for cracked tracks),
etc.), I defrag both the disk(s) and the registry (if applicable).  I
then baseline system performance using any number of software tools to
measure CPU speed, disk and memory R/W performance, etc.).  Sisoft
Sandra is a nice tool for doing a full system check if you don't get too
buried in the numbers. :)

Step 2 (1/2 hour) - Then I look at electrical/electronics issues that
may cause the issue (unregulated voltages, dirty binary signals, bad
drive signals, weak amplifiers, etc.).  When approached logically, the
checks are one of the least time consuming troubleshooting procedures of
any on the list - usually taking no more than 30 minutes.  The "scorched
earth" approach does not solve hardware issues and leaves the customer
with less than what they started with (all of that "old" software).

Step 3 (1 hour max!) - Begin unloading drivers and disconnecting
peripherals and network connections that might impact machine
performance.  If you keep in touch with the manufacturers, this can be a
first step if you come prepared with information that leads you directly
to the problem (conflicts, version incompatibilities, etc.).

Step 4 - (1 hour max!) - Intuition time.  In the few cases where none of
the above solutions work, then it's time to be creative.  This is where
all of those years of banging your head against the keyboard pay off
(head banging is not step 5).  If you've never dealt with a high
resistance fuse then you've never lived!

Notice that the full estimated time is about 30 minutes less than the
estimated scorched earth method that you propose.  I've had my nose
buried in data crunchers since punched cards, teletypes and IBM 360
mainframes and I have never had to take the scorched earth approach on a
bootable system outside of severe data corruption or unrecoverable virus
damage.  I find this topic VERY interesting and would also like to hear
from the others.

        Cheers,
        Steve


-----Original Message-----
From: PCBUILD - Personal Computer Hardware discussion List
[mailto:[log in to unmask]] On Behalf Of Brian Williamson
Sent: April 23, 2002 3:57 PM
To: [log in to unmask]
Subject: Re: [PCBUILD] Customer wants less-sluggish Win9x computer...


I'd be interested in seeing one of these "standards" sheets.  Unlike
some of the other posts on this topic, I usually consider the "scorched
earth" approach early on in the diagnosis process.

                         PCBUILD's List Owners:
                      Bob Wright<[log in to unmask]>
                       Drew Dunn<[log in to unmask]>

ATOM RSS1 RSS2