Why killed "sh NVIDIA-Linux-x86_64-169.09-pkg2.run --uninstall" my 8800GT?

[MagnusBerg]

Yesterday I did a "sh NVIDIA-Linux-x86_64-169.09-pkg2.run --uninstall" while X was running. After that the /usr/lib/xorg/modules/extensions/libglx.so (Debian path) was removed the screen went black and my computer died.
Now I can't start it att all. The led-light lights up and the fans moves for less than a second then i hit the power button. But if i remove the Gainward 8800GT card the computer seams to boot as usual. But I have no other graphic card to test with.
Now I assume that the Gainward 8800GT card has been destroid. But I can't really understand way removing the drivers could cause that. Wasn't the driver loaded into a memory somethere? When I did the "sh NVIDIA-Linux-x86_64-169.09-pkg2.run --uninstall" I expected that nothing att all should happened. The most serious thing I could foresee was that I killed X and land up in terminale mode.

Can someone give me a simple technical explaination how this disaster could happen then I "only" removed the drivers from the hard disc?

[elclanrs]

Try liberating all static energy from the components by plugging off the master cable and waiting a couple of minutes.

I don't think uninstalling the drivers could do that!

[piotrq__]

Ummm... Coincidence? Check the graphics card in another computer to see if it's OK, if not, then RMA it. Just happens.

[MagnusBerg]

I think that elclanrs come quite close with his/her solution...

I called my mail-order firm, komplett.se, and their support was friendly and helpful as always and gave my some suggestions that I chould try to test. One of them was to remove one of the memory modules. And now the computer started up. Then I installed the memory module again, just in case ;-), and to my surprise the computer still started up. Now I have used the computer hole the day and I can't found any faulty behavior. No hardware seems to have been damaged.

That's really happened then the computer went down will remain a mystery...

[xorbe]

Probably a marginal hardware contact.

Run "memtest86" to check memory overnight.