I used to have a great stable system, when a keyboard short caused a huuuge crash. Now, several weeks later it is functioning about 85% as well as it used to.
My biggest problem is still frequent IE crashes. I went back from 5.5 to 5.0 (shipped with 98SE) and that has helped somewhat, but they still occur with ridiculous frequency. However the same crashes sometimes occur in other programs too.
All the crashes are basically GPFs causing page faults in Kernel32.dll, Wininet.dll, mfc42.dll, which from what I have found elsewhere are key files in the GUI shell.
I have searched the MS knowledge base, implemented some of the 'fixes', repaired IE 5.5 several times before going back to 5.0, and re-installed windows a couple of times!
I have tried regclean.exe and Fix-it Utilities 3.0 a couple of times - the last time; Fix-it made over 500 changes to the registry (!!!) causing more than a few failed starts before a windows re-install. I pretty much knew it was gonna happen, but had so little left to lose thought it was worth a shot... Generally, registry cleaning has made things worse not better.
Which leads me to question #1; Can the registry get so hopelessly bogged with changes that it ceases to function correctly? I have been avoiding a re-format because of the huge work that entails and the myriad customizations, drivers etc that are so hard to get back again. I have a LOT of stuff on-board since this is a work unit - about 5GB in programs which are on a separate drive but many of which, these days, seem to require re-installation after a new OS.
Scandisk and Disk Fixer (Fix-it) regularly turn up dish errors (mostly x-linked files, no bad sectors) that probably just come from the stack faults. My anti-virus says I have none.
So, how do you deal with persistent re-occuring crashes? SInce all the hardware was fine before, I'm pretty sure its not a hardware issue. after a couple of re-installs I suppose drivers are a possibility. More and more it just looks like terminal Registry-bloat.
ANY ideas certainly welcome!
Athlon Classic 800
o/c to 950 or running at default doesn't seem to change the error rate
Abit KA-7
128MB RAM
98SE
My biggest problem is still frequent IE crashes. I went back from 5.5 to 5.0 (shipped with 98SE) and that has helped somewhat, but they still occur with ridiculous frequency. However the same crashes sometimes occur in other programs too.
All the crashes are basically GPFs causing page faults in Kernel32.dll, Wininet.dll, mfc42.dll, which from what I have found elsewhere are key files in the GUI shell.
I have searched the MS knowledge base, implemented some of the 'fixes', repaired IE 5.5 several times before going back to 5.0, and re-installed windows a couple of times!
I have tried regclean.exe and Fix-it Utilities 3.0 a couple of times - the last time; Fix-it made over 500 changes to the registry (!!!) causing more than a few failed starts before a windows re-install. I pretty much knew it was gonna happen, but had so little left to lose thought it was worth a shot... Generally, registry cleaning has made things worse not better.
Which leads me to question #1; Can the registry get so hopelessly bogged with changes that it ceases to function correctly? I have been avoiding a re-format because of the huge work that entails and the myriad customizations, drivers etc that are so hard to get back again. I have a LOT of stuff on-board since this is a work unit - about 5GB in programs which are on a separate drive but many of which, these days, seem to require re-installation after a new OS.
Scandisk and Disk Fixer (Fix-it) regularly turn up dish errors (mostly x-linked files, no bad sectors) that probably just come from the stack faults. My anti-virus says I have none.
So, how do you deal with persistent re-occuring crashes? SInce all the hardware was fine before, I'm pretty sure its not a hardware issue. after a couple of re-installs I suppose drivers are a possibility. More and more it just looks like terminal Registry-bloat.
ANY ideas certainly welcome!
Athlon Classic 800
o/c to 950 or running at default doesn't seem to change the error rate
Abit KA-7
128MB RAM
98SE