BIOS Problem: Bootup to Blank Screen

I had a working computer.  Then I decided to fix it.  Now the screen was completely black.  The computer seemed to have booted up nonetheless -- the hard drive light was flashing now and then, suggesting that some program was playing with itself in what I hoped was a nondestructive fashion -- but I could not see anything onscreen.  The monitor was plugged in and turned on, but I guess we were no longer on speaking terms.

What I had tried to fix was a setting in the BIOS.  I was using a Gigabyte GA-MA785GM-US2H motherboard with an Award BIOS (v. 6.00PG).  On bootup, I hit DEL and went into the BIOS settings -- specifically, into Advanced BIOS Features > IGX Configuration > UMA Frame Buffer Size.  My objective was to dedicate some system RAM to video.  So I hit Enter and changed the IGX Configuration from Auto (the default) to 512MB and rebooted.  This gave me the aforementioned black screen, just as it had done for another poor soul.

So now the question was how to fix it.  I first tried to do it blindly.  I booted the machine and kept hitting DEL for a while, figuring that this would take me into the BIOS setup.  Then, following the sequence of steps that would have been required to change the IGX Configuration back to Auto on another machine, I went through a series of keystrokes (Down, Enter, etc.).  Those steps, done in the proper order, took me to the IGX Configuration part of the BIOS on that other machine.  But they didn't seem to work on the blacked-out machine.  When I hit the keys needed to save the settings and reboot, I found myself still looking at a black screen.

If the BIOS was fubared such as to produce a black screen immediately upon bootup, without ever showing a trace of life, then it wouldn't seem to matter whether I booted with a CD, USB drive, floppy, or hard drive.  The one exception, I figured, would be if I booted with some program designed to speak directly to the BIOS.  And for that, the candidate was presumably a BIOS flasher.

In other words, I saw an opportunity, here, to update my BIOS while fixing it.  For this solution, I went to the motherboard's BIOS upgrade download webpage.  Gigabyte had a program called Xpress Recovery2, but its purpose seemed to be to recover hard drive data, not to recover the BIOS.  They also had @BIOS, a live update utility, which would have been great if I could have booted Windows to run it.  The motherboard's manual seemed to be telling me that I needed, instead, to use its Q-Flash utility.  Q-Flash was said to be embedded in the motherboard's hardware, so I wouldn't need any particular drive to run it.  It said I could use Q-Flash to install a BIOS update that I would download on the other computer and save to a FAT32/16/12 USB flash drive.

But could I use Q-Flash if I couldn't even see it?  The way to fire up Q-Flash, according to the manual, was to hit the End key while the system was booting.  The blanked system was currently running, so I tried using WinKey-U-R to restart it, since I could see that those keystrokes were what it would have taken to reboot the other computer from Windows 7.  I gave that several minutes, since I had no idea what was running on that computer at this point.  I never got a beep, though, so I thought maybe it was waiting for me to Force Restart.  I hit the F key.  Nothing happened.  I tried Enter.  A brief hard drive flicker.  I gave it another minute and then just punched the reset button on the computer.  Then I kept hitting End for a while.  Perhaps I was now in Q-Flash.  No way of knowing:  the screen was still blank.

I thought of trying to trace my way through a BIOS flash blindly, as I had tried to trace through the reset of the IGX Configuration option.  Thinking of that gave me an obvious idea:  reset!  Maybe I could just take the steps needed to reset the entire BIOS back to its defaults.  I punched the computer's restart button again and then, after I got the reboot beep, I kept hitting Del, twice a second for about 15 seconds.  The keys I hit at this point (copied, again, from the sequence on another machine running a hopefully similar CMOS setup utility) were:  right-arrow (to take me to the option for Load Fail-Safe Defaults), Enter (to actually load those defaults), then Y to confirm, then F10 to save and exit.  That produced no results, so I hit Esc several times, in hopes of backing out to the main CMOS menu, and tried again:  Right, Enter, Y, F10.  This time I added another Enter for good measure.  And oh, my Christ, it worked.  I was able to read my screen again.  Fricking brilliant.  Amazing what you can do when you can't see a thing.

I went back into the BIOS, because of course I hadn't had enough of this, to take a look at how things were now.  The UMA Frame Buffer Size was back to Auto.  Funny, I didn't recall even seeing a UMA Frame Buffer option on the other computer.

It seemd obvious, now, that I should have just gone directly for the Fail-Safe option in the first place.  I reconfigured the BIOS as desired, saved, and rebooted.  Everything was fine.  I wasn't going to need to root around anymore in my Google search for solutions.  Although I did realize, a bit later, that I probably could have achieved the same thing, without working blindly -- resetting the BIOS (a/k/a clearing the CMOS) -- by either removing the quarter-sized battery from the motherboard for five minutes or shorting across the motherboard's "clear CMOS" jumper, which the manual would probably have helped me to find.

But no.  Not so fast.  On reboot, I was back to a black screen.  Why?  I hadn't even touched the IGX Configuration stuff this time around.  But, ah, false alarm.  Apparently the fail-safe options concealed the Power-On Self-Test (POST) information.  After a short panic, I had Windows onscreen.  I'd just have to take another look at the CMOS setup, next time I rebooted, to find the setting that would restore the POST display during bootup.  There may have been a way to do that with Gigabyte's Easy Tune utility, though if there was, it wasn't immediately obvious to me.

I decided to go ahead and deal with that now.  Unfortunately, when I rebooted and hit Del repeatedly, it just gave me a blank screen.  I hit Esc and then Enter, to exit the BIOS and reboot without saving any changes, but that didn't do anything.  I tried again, and then tried F10 and Enter.  After a blank screen, that got me back into Windows, at least.

Well.  Were the fail-safe defaults preventing me from getting into the CMOS setup?  It seemed that maybe I should go ahead and update the BIOS after all, or else open the computer and use one of those hardware BIOS-reset methods.  I ran Gigabyte's @BIOS utility and selected the "Update BIOS from Gigabyte Server" option.  I had to approve a couple of choices, and then it ran.  In a half-minute or so, it had apparently downloaded the new version.  It said this:

The screen will freeze for a few seconds while updating the BIOS.
Do you want to update the BIOS?

I clicked OK.  After a moment, it said, "BIOS Update completed!  You must restart your system to take new changes."  I said, "Restart Later."  I didn't want to lose all the stuff I had open, so I hibernated the machine (Start > Shut Down > Hibernate) and then, after it died, I pushed the power button and started it back up.  That worked:  I could now see the POST screen.  I hit Del and went into the CMOS setup.  I had to reset the clock and make other adjustments.  Then I rebooted.  And yet, once again, I was not seeing the POST screen, though once again at least the computer did proceed on into Windows.  It seemed that maybe one of my changes was responsible for this, or else perhaps that the hibernation was fouling things up.

It was hard to tell what ultimately fixed this.  Something did.  When I returned to these notes a while later to wrap up this post, I was no longer having the problem.  Possibly the steps described here did solve it on reboot, though I think in that case I would have made note of it.  It seemed I would need to have the problem again in order to comment further on it.

Documenting Computer Work Onscreen - Second Try

I had previously used my Shotshooter.bat batch file to capture screenshots once or more per second, and had combined them into an IrfanView slideshow .exe to display a timelapse account of what I had been doing on the computer during a certain period of time.  I preferred to create a regular video file rather than an executable, and I also wanted to be able to add music, narration, or other audio.  So I tried again.

The problem wasn't video capture per se.  I had purchased a copy of Debut for that, after mixed results with freeware alternatives.  The problem was getting from the raw Debut output to the final video.  When I would import the Debut .avi into a video editor like Adobe Premiere Elements or CyberLink PowerDirector, the video quality would be seriously degraded.  It was possibly a problem with a simple setting in the editing program, but I hadn't yet figured out the solution.  At this point, for whatever reason, the resulting video of events on my computer's monitor was so poor as to be unreadable.

But let me back up.  Here's how I started.  I decided to try capturing my work in writing an essay.  It would wind up being a one-page document, so the whole thing was visible onscreen in Microsoft Word.  I had Word set to remember changes, so after hours of editing, I held down the Ctrl-Z key combination to keep undoing changes until I was back at the beginning.  (Before doing that, I saved a copy of the text.  One false keystroke would have destroyed the trail, and my essay would have been lost.)  When I was back at the beginning, I started the screen capture, using Debut.  Then, in Word, I held down the Ctrl-Y key combination to redo everything I had done, at a very rapid pace, while Debut was capturing.  I think I had set Debut to capture at seven frames per second (fps), so as to keep the .avi file somewhat smaller than it would have been at 30 fps.  When it was done, I stopped Debut and saved the .avi.

Now came the hard part.  Adobe and CyberLink weren't the only ones having problems with the Debut .avi output.  Media Player Classic (MPC) would likewise display that .avi in a severely distorted form.  In other trials, MPC played the audio but gave me only a black screen for the video.  QuickTime did just the opposite, playing the video but crapping out on the audio after a few seconds.  But for some reason, VLC and Microsoft's Windows Media Player (WMP) had no problem with it.  It was not clear why.  I would have guessed that my codecs needed updating, but I had just installed the latest K-Lite Mega Pack.

After much playing around, I found that VirtualDub was able to add an audio track, adjust the length of the video to the length of that audio track, and produce a relatively small .avi (but, regrettably, not an .mpg or some other compressed format) that WMP would play.  This video was about four minutes long and, at a 250 kbps setting (Microsoft Video 1, quality 100), it was about 70MB.  That was still huge in comparison to the IrfanView output, but it was nice to have the audio option.  At that bitrate, the video contained fade-like artifacts.  That is, when I deleted or changed some text in the original video, it faded out rather than just instantly disappearing.  It actually wasn't a bad effect, and I didn't want to upload a 700MB video, so I left it at that for now.

There seemed to be some artistic possibilities for this technique.  At some points, the beat of the music would coincide with the disappearance of lines of text, as if there were a rhythm to the editing.  Maybe there was, in some cosmic sense.

The remaining question was whether anyone else would be able to view this video.  I wasn't sure how it would turn out on YouTube.  So I uploaded and tried it.  It looked OK in full-screen mode on YouTube, viewed in Firefox and also in Chrome.  I downloaded it from YouTube, using NetVideoHunter.  The download was in .mp4 format.  It still didn't play right in WMP, but it played better in QuickTime, and it was less than half the size of the .avi I had uploaded.  So I thought perhaps I could have uploaded the 700MB version and used YouTube to compress it for me.

I still didn't have a straightforward process for putting high-quality screen capture video into a format that I could edit in a normal video editing program like Adobe Premiere Elements.  It seemed that, next time, I might try recording my original video in something other than Debut, to explore the possibility that that particular program had saved my video in an odd format that Premiere Elements and other programs couldn't handle.  I could also explore rapid Shotshooter screenshots, at least to the extent that my computer would be able to save multiple screenshots per second.  Then maybe I could also try using Premiere Elements, instead of IrfanView, to stitch those screenshots together into a time-compressed but visually high-quality production to which I could also add audio.

Windows XP in VMware Workstation 7.1: Windows Explorer Keeps Refreshing

I was running Windows XP SP3 as a guest in a virtual machine (VM) in VMware Workstation 7.1.1 on an Ubuntu 10.04 (Lucid Lynx) host.  I had just installed a couple of freeware utilities when suddenly Windows Explorer began refreshing itself about every two seconds.  More specifically, it was refreshing the right-hand pane, showing files and folders, but not the left pane, showing the folder tree.

I thought at first it was a virus, but I was running antivirus software, and a scan with a different antivirus program turned up nothing.  Besides, I had just downloaded those programs from reputable sources (e.g., CNET) that supposedly certified them to be virus-free.

A search suggested this was a relatively common problem.  Several posts made me think it had to do with network drives, which in this case would mean the link between Windows and VMware.  I killed the VM, reverted to a previous snapshot, and started that, but the same thing was happening there as well.  I powered up a different VM in a different session of Workstation.  The problem was not occurring there.  I closed all sessions of Workstation and powered up the misbehaving VM in a new session of Workstation.  The problem recurred -- but then, after a minute or two, it stopped.  But then, after I used Windows Explorer some more, it resumed; but then it stopped again.

I downloaded the "Prevent Automatic Folder and Icon Refresh" registry edit from Kelly's Korner and ran that, and then rebooted Windows within the VM.  The problem was still there.  I took this to mean that *automatic* refresh was not the problem -- that, presumably, something was manually refreshing Explorer.

I tried displaying different drives' contents in Explorer, on the theory that maybe this was happening in connection with just one particular network drive.  That was not the case; it happened on all of them.

I wondered if a program installation was responsible for the problem.  But it was not clear to me how any of these would have been responsible for the fact that the flashing occurred within a snapshot that I had taken before installing them.

Then I came closer to what seemed like a possible answer.  I tried again to install the Microsoft Task Switch Powertoy.  I had tried before, and it had given me an error message, and the same thing happened again.  The error message was "Error 1606.  Could not access network location [gibberish]."  The gibberish was actually just a set of five squares with no characters in them.  It appeared that the installer was trying to access a network location whose name did not consist of valid characters.  Following advice, I checked the following registry keys for incorrect addresses:

HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\Shell Folders
HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\User Shell Folders
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\Shell Folders
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\User Shell Folders

I did find several bad entries in the second of those five locations.  I corrected those and rebooted the VM.  I was now able to install the PowerToy.  Unfortunately, the flashing was still going on.  I went into VM > Settings > Options > Shared Folders and switched that to Disabled > Save.  The flashing stopped.  In Windows Explorer, I tried to go to another network drive, but got the message that it "is not accessible.  The network path was not found."  I went to drive C (not a network drive).  Its contents displayed OK.  No flashing.  I set Shared Folders back to Always Enabled.  No flashing on drive C.  Flashing on drive D.  I tried C again.  The folders there would refresh once, immediately after being selected, but then not again.  In Windows Explorer's menu, I went to Tools > Disconnect Network Drive and disconnected drive D.  Drive E was still flashing.  I went to Tools > Map Network Drive and mapped D again.  It was flashing.

A Microsoft Knowledgebase webpage said that a somewhat related problem (flickering in the left-hand pane of Windows Explorer) could be repaired at HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\Explorer.  I had only a lowercase "policies" (not "Policies") key at that location.  Regedit would not let me create the uppercase version.  So I went into the lowercase "policies" key.  It did have an Explorer subkey.  I went in there and created a new key, NoRemoteRecursiveEvents, of REG_DWORD type, and gave it a value of 1.  I exited regedit and rebooted.  This step did not solve the problem.

I closed this VM and went back to the previous version, the one that did not flash.  I corrected the incorrect registry addresses (above) first.  Then I started reinstalling the programs that I had been installing when the problem began.  I went down the list and, what do you know, the flickering started when I installed FolderSize, and it stopped when I uninstalled it.  FolderSize calculates the size of folders, and apparently refreshes the screen while it does so.  After I uninstalled it, the problem went away.