Recently in Hardware Category

Both ATI and NVIDIA have published newly-signed WHQL drivers for their respective lines of graphics cards; my own ATI (a Radeon 9550) got new drivers just this morning courtesy of Microsoft Update.  I'm not sure about this -- it may simply be some observational bias on my part -- but some parts of the system seem to be running a little bit faster with the new driver.

However, the big thing for me is and will probably always be stability.  Microsoft moved the video drivers out of kernel space to sacrifice that much more speed for that much more stability -- i.e., they moved back to the old NT 4.0 way of doing things -- and from what I can tell the tradeoff hasn't been anything that's really hurt me.  (The FPS hounds, though, may tell a different story with Vista -- but I'm admittedly not a gamer, so this whole area of discussion is really outside my realm of expertise.)

I feel a little foolish for not taking picture of this earlier, but when I was having my issues with my other video card, I checked out the Reliability and Performance monitor just to get an idea of how those crashes were affecting the system stability rating.  The chart was nosediving like the stock market after a heavy selloff.

Now, after switching back to the ATI card and a stable driver, here's what it looks like:

I had a hard time not feeling smug.

The problems I was having with Vista and my NVIDIA video cards appear to be a known factor.

The problems with the early ATI drivers on Vista were problematic enough to make me believe the card itself was defective: the system would freeze, hard, with no crash dump and no emergency video recovery (one of the neat new things about Vista's graphics subsystem).  The NVIDIA GeForce card ran, but slowly, and every other time I rebooted it would BSOD.  Eventually it would BSOD every time I rebooted, and finally just stopped booting altogether.  I was tempted to believe any number of things were wrong, up to and including the possibility that the video card simply didn't like multi-processor systems.

After new ATI drivers arrived directly from Microsoft, I switched back to ATI and haven't had a problem since.

Another of those infamous random freezes killed my computer this morning.  This was after I'd spent all night running Memtest86+ to make sure I wasn't dealing with a rogue bad-memory issue; I got through eight passes with the program (with cache both on and off) and found no problems.

Whenever I've dealt with a problem like this in the past, my usual way of debugging it has been to first remove everything that does not absolutely have to be there, and then cycle out the components that can be changed.  Since I no longer have any expansion cards left in the system, I decided to cycle out the video card (an ATI Radeon 9550) and replace it with an older one that had originally shipped with the PC (an NVIDIA Quadro FX 500).  That meant losing Aero for the time being, but I'm patient and losing Aero for a bit is not going to break me.

I'm now tentatively wondering if the Radeon a) had some kind of "sleeper" issue that was only exposed when Vista started really stressing it a bit, or b) it was already starting to malfunction and it did so coincidentally with my move to Vista.  It's a bit of a parallel with the folks who moved to Vista and then discovered one of the DIMMs in the PC in question was bad -- and that this only surfaced with Vista because of the ASLR (address space layout randomization) system in use.  Likewise, maybe there was a defect with the card that only surfaced in the rarest circumstances before, but is now surfacing more regularly with Vista due to differences in the way the card's being handled.  It's tempting to blame Vista, but off-base; it doesn't solve anything.

If I go for three days without a crash using this older card, I'll swap the other one back in and try it without Aero to see if that makes any difference.  Either way, I think I may be looking at a new video card.

One last possibility is that I might be looking at some weird side effect of DEP.  If I continue having problems, I might try turning DEP off (or put the old card back in and try it with DEP off to see if that affects anything).  A shot in the dark, but a shot worth taking.

Remember those random Vista freezes?  One of them struck last night in the middle of work -- in fact, it trashed a document that I was in the middle of working on (thank you, Word Auto-Save), and I had to take a picture of the screen to recover the part of the document that I had been most immediately typing.

So far I'm going with one of a few theories:

1. A driver issue.  I've upgraded all of the drivers I can in this system to their most recent editions.  The one thing that I tried most recently after this last freeze was updating the video driver to ATI's Catalyst-edition drivers (as opposed to the bare drivers I get from Windows Update).  I've heard that the Catalyst software is a little more thorough about spec-ing video card settings that might cause instability, so I'll try it.
2. A memory issue.  All four 512MB DIMMs in this machine are registered ECC, though, and I have memory error correction turned on in BIOS; so I'm not sure how likely this is.
3. An interaction with another component.  The freezes didn't stop when I pulled out the one other expansion card I have in the system -- a fairly generic Belkin PCI USB 2.0 card -- so if it's something else that's in the system it might take a lot more hunt-and-peck to find it.

Also, I managed to trigger Product Activation during this experimenting -- probably when I yanked out one, then the other set of DIMMs to see if that fixed the instability (it didn't).  But I put everything back to the way it was and re-activated.  I wonder if there's a way to back up and restore Product Activation in Vista as there was in XP.