Thursday, September 11, 2008
Fantastic Microsoft Idiocy
Some time yesterday, there was a power outage at my place of sufficient length to cause my UPS software to hibernate the machine. That apparently went OK, but when the machine tried to restore from the hibernate file, it instead bluescreened and rebooted.
This resulted in a familiar screen to frequent hibernate users to appear upon rebooting. It's the page of text that tells you that the last attempt to restore from hibernation failed, and asks if you want to continue with system restarting or delete the hibernate file and then start the system.
That's all fine and good. In cases like this, I usually want to try one more time to restore from hibernation (which works about 20% of the time), and when that fails, I just cut my losses, delete the hibernate file and continue.
The problem? This menu does not work, at all, if you have a USB keyboard. At first I thought it might be a problem with my KVM switch, so I took that out of the loop and connected my USB keyboard directly to the PC. No luck. Then I thought maybe there was a problem with my new USB keyboard, so I tried accessing the BIOS with it. That worked fine.
Fortunately I still had my old PS2 keyboard laying around, and I could plug it in to respond to that prompt. But anyone without a PS2 keyboard handy would be unable to boot into windows until they got one.
Somehow, Microsoft, one of the largest software companies in the world, with its thousands of engineers working on Windows XP, can't manage to make a simple menu work with standard USB keyboards. Idiots.
Postscript: It seems a number of others have gotten into this particular failure mode. A Google search for windows hibernate "restoration data" "usb keyboard" turns up a number of people experiencing the same problem, and mostly solving it the same way: by plugging in a PS/2 keyboard to respond to the menu.
This resulted in a familiar screen to frequent hibernate users to appear upon rebooting. It's the page of text that tells you that the last attempt to restore from hibernation failed, and asks if you want to continue with system restarting or delete the hibernate file and then start the system.
That's all fine and good. In cases like this, I usually want to try one more time to restore from hibernation (which works about 20% of the time), and when that fails, I just cut my losses, delete the hibernate file and continue.
The problem? This menu does not work, at all, if you have a USB keyboard. At first I thought it might be a problem with my KVM switch, so I took that out of the loop and connected my USB keyboard directly to the PC. No luck. Then I thought maybe there was a problem with my new USB keyboard, so I tried accessing the BIOS with it. That worked fine.
Fortunately I still had my old PS2 keyboard laying around, and I could plug it in to respond to that prompt. But anyone without a PS2 keyboard handy would be unable to boot into windows until they got one.
Somehow, Microsoft, one of the largest software companies in the world, with its thousands of engineers working on Windows XP, can't manage to make a simple menu work with standard USB keyboards. Idiots.
Postscript: It seems a number of others have gotten into this particular failure mode. A Google search for windows hibernate "restoration data" "usb keyboard" turns up a number of people experiencing the same problem, and mostly solving it the same way: by plugging in a PS/2 keyboard to respond to the menu.
Labels: microsoft, shitware, stupid, windows
Monday, July 14, 2008
Solved: the monitor/video card ass-pain
I am not a gamer. Let's just get that out of the way right now. With my desktop PC I do things like edit sound files with Sound Forge or browse the web or edit documents. My video needs are minimal. The biggest workout my video card ever gets is playing video files with VLC.
I am, however, a bit of an energy efficiency nut, which is why when I designed my "new" PC (now a couple years old) I chose a reasonably capable motherboard that had decent onboard video and sound and minimal power consumption. While I did eventually end up upgrading the sound card (the one on the board ultimately ended up being too noisy and always required DC offset adjustment), I have never had any desire to upgrade to an energy-hogging video card with 8 fans and its own power connector. I like my cool, 100-watt, quiet PC as it is.
Almost the entire time I've had this PC, however, I've been plagued by an irritating problem. If I turned off the Viewsonic vx2035wm monitor I had plugged into the Wintec ADD2 DVI card (one thing the Intel DG965OT board lacked was an on-board DVI connector) and turned it back on, the resolution changed from 1680x1050 to 1400x1050. Sometimes Windows would also change its idea of my monitor from the Viewsonic to the Default Monitor. No amount of BIOS upgrading or video driver updating or monitor driver updating seemed to make any difference.
Today I decided I wanted to rule out a problem with the Wintec ADD2 card, so I pulled the card and plugged the monitor into the analog VGA port. Of course this makes the video look quite ugly. What I found was that after installing the drivers for the monitor, it wasn't getting detected, and if I updated the driver to the Viewsonic one by hand, I wasn't able to even select the native resolution at all.
That led me to Intel's solution ID CS-028366. A problem with reading the EDID from some flat panel monitors with Intel's G965 video chipset leads the driver to fail to see the native resolution of the display. Intel cooked up a special driver for people experiencing this problem, which can be downloaded from the solution page. I updated to this driver, and that made it possible to select the native resolution for the display in analog mode.
This got me to wondering if misreading the EDID could cause the problem I had with the digital connector as well, so I powered down and reconnected the ADD2 card. After restarting I found that in fact, the native resolution of the monitor was selected each time if I turned off my monitor and turned it back on now. Problem solved, provided it doesn't randomly return as things sometimes do in the Windows XP world.
Intel doesn't provide much in the way of technical details about what causes the problem or what their fixed driver actually does, but we might be able to guess from their saying "there may be side effects to this solution such as longer than normal resume times from standby" that either they're requesting the EDID at a lower bit rate, they're requesting the EDID multiple times looking for matching data, or they're throwing in a short delay after waking up the monitor and before requesting the EDID (perhaps monitors with limited horsepower can only tackle one task at a time and the delay permits the monitor some time to finish starting up the display before having to answer an EDID request).
In any case, I'm glad I can finally turn my monitor off.
I am, however, a bit of an energy efficiency nut, which is why when I designed my "new" PC (now a couple years old) I chose a reasonably capable motherboard that had decent onboard video and sound and minimal power consumption. While I did eventually end up upgrading the sound card (the one on the board ultimately ended up being too noisy and always required DC offset adjustment), I have never had any desire to upgrade to an energy-hogging video card with 8 fans and its own power connector. I like my cool, 100-watt, quiet PC as it is.
Almost the entire time I've had this PC, however, I've been plagued by an irritating problem. If I turned off the Viewsonic vx2035wm monitor I had plugged into the Wintec ADD2 DVI card (one thing the Intel DG965OT board lacked was an on-board DVI connector) and turned it back on, the resolution changed from 1680x1050 to 1400x1050. Sometimes Windows would also change its idea of my monitor from the Viewsonic to the Default Monitor. No amount of BIOS upgrading or video driver updating or monitor driver updating seemed to make any difference.
Today I decided I wanted to rule out a problem with the Wintec ADD2 card, so I pulled the card and plugged the monitor into the analog VGA port. Of course this makes the video look quite ugly. What I found was that after installing the drivers for the monitor, it wasn't getting detected, and if I updated the driver to the Viewsonic one by hand, I wasn't able to even select the native resolution at all.
That led me to Intel's solution ID CS-028366. A problem with reading the EDID from some flat panel monitors with Intel's G965 video chipset leads the driver to fail to see the native resolution of the display. Intel cooked up a special driver for people experiencing this problem, which can be downloaded from the solution page. I updated to this driver, and that made it possible to select the native resolution for the display in analog mode.
This got me to wondering if misreading the EDID could cause the problem I had with the digital connector as well, so I powered down and reconnected the ADD2 card. After restarting I found that in fact, the native resolution of the monitor was selected each time if I turned off my monitor and turned it back on now. Problem solved, provided it doesn't randomly return as things sometimes do in the Windows XP world.
Intel doesn't provide much in the way of technical details about what causes the problem or what their fixed driver actually does, but we might be able to guess from their saying "there may be side effects to this solution such as longer than normal resume times from standby" that either they're requesting the EDID at a lower bit rate, they're requesting the EDID multiple times looking for matching data, or they're throwing in a short delay after waking up the monitor and before requesting the EDID (perhaps monitors with limited horsepower can only tackle one task at a time and the delay permits the monitor some time to finish starting up the display before having to answer an EDID request).
In any case, I'm glad I can finally turn my monitor off.
Labels: annoyances, hardware, windows
Monday, April 28, 2008
Fixing that Annoying Windows XP Taskbar Resizing Problem
Windows XP can be an annoying POS sometimes. Especially when it thinks it knows better than I do how many lines my taskbar should be. I like it to be 2 rows high. For reasons unknown, Windows wants it to be one row less than whatever I set it to. Every time I reboot, Windows reduces the height of the taskbar by one row of buttons.
Or it did. I made it stop. It seems like this is related in some way to Windows' theme support. One way to make it stop is to disable that. In the Control Panel, open Administrative Tools | Services. Look in the list for "Themes". Right-click it and choose "Properties." Change the Startup Type to "Disabled" and click "Stop."
Now Windows XP doesn't change my taskbar anymore. And before you ask, yes I tried locking the taskbar first. It had no effect, other than making it take one more step for me to resize the taskbar back to where I set it.
Or it did. I made it stop. It seems like this is related in some way to Windows' theme support. One way to make it stop is to disable that. In the Control Panel, open Administrative Tools | Services. Look in the list for "Themes". Right-click it and choose "Properties." Change the Startup Type to "Disabled" and click "Stop."
Now Windows XP doesn't change my taskbar anymore. And before you ask, yes I tried locking the taskbar first. It had no effect, other than making it take one more step for me to resize the taskbar back to where I set it.
Subscribe to Posts [Atom]
