Level Extreme platform
Subscription
Corporate profile
Products & Services
Support
Legal
Français
MAJOR problem with my PC - UPDATED
Message
From
10/01/2007 06:51:46
 
 
To
09/01/2007 17:02:55
General information
Forum:
Windows
Category:
Computing in general
Miscellaneous
Thread ID:
01183991
Message ID:
01184119
Views:
18
My sympathies Jim...
>Well I just called the shop after waiting patiently for 1.75 days.

From what you posted they are not top of the line.

>Seems their tests are indicating memory (access or addressing) problems and the issue persists even with replacement RAM installed.
>So the prime suspects now are:
>1) motherboard L2 cache (or something else on MB);

No MoBo Cache on any AMD K7 and above architecture at all. Perhaps some interrupt mixup on the busses or the chip responsible are fritzed.

>2) CPU cache (an AMD X2 - 4400+ socket 939 installed new within 6 months).

Most BIOS can disable L2 cache, and switchinga CPU for test purposes should be done in a few minutes
(unless it is one of those cramped "small" cases)

>
>Here's the hooker...
>The motherboard is discontinued and cannot be ordered (ASUS A8V Deluxe) so I'm told that if it turns out to be the motherboard it has to go through the "RMA" process which sounds like it could take weeks!
>They seem to be oblivious that it is my WORK SYSTEM.


>
>I asked about an equivalent replacement and they are hedging on the issue.
>Their main concern is that the RAIDed C: will be clobbered by any change in controller (I used the Promise controller on the ASUS (it has 2, VIA as well)).

Order of testing: Put the drives in another system with a promise controller, boot from other disk check from that boot (no need to write into the partition). If that works, save data and try booting on that system. Then test your old system with an exchange CPU or get the CPU from your old system,

If the above doesn't work, you don't want to keep the raided C: IAC. Test the old system with another disk to boot from, test you can access your data partitions, work on and later recover/unRAID your previous C: partition.

>Lessons learned:
>
>1) WHen buying a new PC, buy an extra motherboard;
IMHO a fallacious analogy here: For my clients with a couple of thousand PC's it makes sense to standardize to an even greater extent, as the few spare systems amount only to a percentage and it is certain some machines will fail. In your case a MoBo might grow old just as a safety precaution - that makes innovating to expensive! Standardize on smaller items, use only features common or have a backup boot if you install system specific chipset drivers. I "standardize" on RAM used, type of VGA chip and slot - if a system "goes", parts will be re-used and that's it. I am even able to switch VGA cards without reinstalling drivers<g>. Don't install anything with a single point of failure (I do have a spare SCSI controller for instance), which in your case translates to buying at least a SATA interface card to SATA enable your other boxes (and thereby staying productive after a disk switch). Or wait with your purchase until you need 2 systems OR a group of 3 needs PC's, thereby giving you redundancy a visit away - which is the case for the X64 DDR/184Pin I use - 4 other machines with identical build nearby so I won't be floored. If you tie yourself to the MoBo by buying 2 in case of the chipset bug in the VIA KT133A you would have had 2 faulty boards.

My take is that your goal should NOT be to guarantee one sepecific machine to run all times by buying spare parts! You should aim for a setup flexible enough to allow you to switch to ***any*** of your other machines within half a day at maximum including OS updates and possible installation of some software. You get a much better redundancy which translates into higher reliability - what happens if the CPU of your double MoBo croakes on a weekend bracketed with other holidays and no other has the same socket ? If you need other functionality, look for USB approaches - instant switch possible. I know you also can think about HW at a system-wise lower level, so get out excel, make a comparison list of the HW you own in each machine and what is missing to avoid single fault paths. Hint: a laptop has more single fault paths than other machines.

>2) Don't bother with RAID, at least for C:. I gets you no benefit and lots of trouble.

My take as well. Even for backup in the range I need (day or week) RAID1 would be bad, as it opens me to the risk of the whole machine being stolen. I use RAID only to speed up temp data.

regards

thomas
Previous
Reply
Map
View

Click here to load this message in the networking platform