Provided the systems that are crashing are stable (MemTestx86+, Prime95 on balance for the proper duration, FurMark etc) I think the best course of action here before an all out debugger attachment, would be to get the CPU-Z (cpuid) output from the affected people, notably the SSE extension capability of the processors in question.
A clear-cut example:
An Athlon64 Winchester, will crash randomly with Titanium eqgame.exe
An Athlon64 Venice, will not
The differences between these processors (other than memory controller clock cycles and TDP) is most notably SSE2 vs SSE3. There is even a frame rate difference between the two processors, clocked at equal speeds.
Another issue I've had with Titanium that pre-dates Project 1999; Both InnerSpace and WinEQ2 under Vista like interfering with eqgame.exe's 3D offsets. The transition between server select (or disconnect text, or zoning render surface) often times grayed out for me (and cavedude at the time) under Windows Vista with nVidia 16x.x drivers, and ATI Catalyst 8.x. If you clicked that grayed out window, the exit code for eqgame would be 0 (operation completed successfully) in a debugger. I suggest anyone with older drivers, update to the most recent version they can find, even if it means adding a deviceID to a driver INF. The issue hasn't revisited me since I switched to Windows 7, and it never existed in Windows XP x64.
|