My AMLI: ACPI BIOS wants to do what?
So it's entirely possible that the crashes I was experiencing this weekend weren't RAM related. Investigating the event log shows that my "ACPI BIOS is attempting to read from an illegal IO port address (0xcfc), which lies in the 0xcf8 - 0xcff protected address range. This could lead to system instability." What's fairly ironic is that Windows, upon noting that my ACPI BIOS is trying to fiddle with things that should not be fiddled with, resets the computer. I suppose that fixes the "instability", though I can't say it does much for usability.
(Which is a cheap jab: if Windows detects any kind of corruption, I'd much prefer that it reset rather than write corrupted memory to, say... the master file table.)
So after installing the RAM, seeing Windows start up fine, proceeding to use it for a while, and then having it crash under a heavy load, I ran memtest86
, which found memory problems... but only in the second stick of 512MB. After more fiddling, I find that when used alone, each stick of 512MB works fine; when paired with my old 256MB stick they work fine together; and so it's only when the 512MB sticks are paired do they conspire against me. So I increase the CAS latency (the sticks claim to have a CAS latency of 3 and unfortunately my mobo only lets me go up to 2.5 - fingers crossed) and things seem to be better. I run memtest86 all Sunday night and Monday day - everything looks good.
Until I'm running under a heavy load with some insane IO going on. It crashes again. My older hard drive has this nasty habit of, when under heavy load, CLANKing loudly as the head smacks home, spinning down, then spinning back up. So far I've seen no ill effects, but I'm always wary of it. The last crash seemed to be preceded by the HD once again beating its head, so I thought maybe it was the old drive failing. Soooo... I power up the ol' Western Digital HD Utils, which lives on a floppy made in 1996 and last written god knows when. How this thing still works
is beyond me, and so I place it in a mental category containing a variety of mystical things that I can't quite explain. It cheerily tells me that there's an "error" on my drive that it may be able to repair. And that's all
it tells me - nothing so boring as the specific error or how, precisely, it intends to repair my system drive. But we all know what telling it to go ahead means, and it hasn't come to that yet.
So I hop on Western Digital's website to check for a more recent version of their HD Utils, which I hoped would be more... descriptive. The utils are now in ISO form, which is perfect, because of the floppies I have laying around, I doubt any are newer than seven years old. The last thing I want is the "Display welcome message" instruction to be secretly switched on the floppy with the "Write random data over the most sensitive regions of both hard drives" instruction.
And, of course, the gods of system debugging smiled on me. My shiny new version of Western Digital Data Lifeguard Diagnostics for DOS version 5.04c showed... absolutely nothing. Clean bill of health. Which I question, but I'll accept over the option of having 1996 attempt to repair my drive.
Allow me an opportunity to gush about Western Digital. Everyone has that one company whose drive reduced itself to rubble as they were printing the final paper for their college capstone course five minutes before it was due. And everyone has that one company whose drives have filled their lives with sunshine. Western Digital hard drives consistently give me both milk and
honey. If you're having problems, you simply click on the "Contact support" tab of their website, where you are directed to no fewer than 17
ways of contacting support. There's an "Ask a Question" form, e-mail addresses for support in five different languages, and phone numbers for eleven countries, including a toll-free
number in the states. Now contrast this with Roxio's tech support, in whose dubious honor I spewed vitriol in yesterday's post
: to download updates I had to register - giving out my e-mail address and the CD key included with the software; I could not reach any means of contacting support because I had to provide the "tech support ID" included in the retail box - which the website indicated was invalid; and the only phone option was for non-technical
support with a phone number in Canada. So yes, as the kids say, Western Digital "r0x0rs my s0x0rs."
Now having determined that the problem was likely not the RAM or the hard drive, I dug into the event viewer and discovered several messages from the USB smart-card reader I use to connect to work. Google tells me that the crashes may be related to the onboard USB... which makes sense because this weekend I also installed the adapter to give me two more ports.
Out of all the crazy things I did to my computer this weekend, the one I thought least likely to be the source of the crashes ends up being the prime suspect. Good times... good times.
As Drudge says, "Developing..." (Honestly, are things ever not
developing? His final post will be, "Dirty liberals finally caused the end of the universe. Developing...")
@ 7:04 PM, March 22, 2005
FWIW, I've found that my desktop dies if I plug in my USB 2.0 6-in-1 memory card reader and do some heavy I/O copying images from my digital camera's CF card. It just hangs, and on reboot I get the following message in the BIOS:
Press ESC to skip m
Yes, that's right, it hangs -- consistently -- between the "m" and "e" in the word "memory". Unplug the USB device, and it all works again...
@ 7:10 AM, October 05, 2007
I would never had thought it would be related to that! just borrowed a card reader yesterday and then decided to reboot... I thought something much worse was happening. thanks a lot for displaying that info !