An observation i had.

[Corlin]

I have an older computer, and would routinely crash after 1) too many hours (say 4-6) of continuous play, regardless of /flush or 2) when I ran across the world chunking 8 or more times in 20 minutes. That always happened during beta, and was much worse on live post release.

I find on test that (1) never happens now, though (2) still does. Full file scan has no effect. Virus checker has no effect. Not sure why server load would matter for CTDing, but it's the only thing left I believe...

[Chantillary]

I have this problem myself. I tend to crash out about every 60-90 minutes. And every time it happens, the same thing is at fault -- memory problems.

Now, my machine has 4Gb of DDR400 and I've reserved 8Gb for virtual memory (and it's all one chunk, unfragmented) and yet the memory bugs happen at ~ 1Gb of physical RAM and 1.25Gb of virtual memory. The VGClient.log file always shows basically the same thing, that the system was unable to write some file or other (normally about 2Mb in size) to VM.