Nothing wrong in the logs... Viewer side, everything seems fine. The watchdog kicks in when the graphics driver fails to return from an OpenGL call in time, indicating a thread crash, a freeze or an infnite loop during the graphics driver execution. You could try and run the graphics driver in non-threaded mode to see if it still crashes and, if it does, whether you can get a better crash report (not involving the watchdog, which can happen in non-threaded mode).
Apparently, the Cool VL Viewer's optimizations fare much better with the new CPU...
The issue is probably with the GPU, not the CPU (the latter would cause a crash in the viewer code, while the former crashes during the graphics driver execution). At this frame rate, your GPU could get close to overheating, or it could stress too much your power supply unit (if it is not powerful enough), or this could be a problem with overclocked CPU/RAM/GPU/VRAM. Watch your cores and GPU temperatures, and see if any is getting too hot.
Note that after building a new system, it's a good practice to stress-test it in order to check its stability under heavy loads. I'd recommend you to run
BOINC with a project such as Einstein@Home and/or Milkyway@Home: these really get both the GPU and CPU hot (you'll probably never get them hotter any time) and 100% loaded (especially with SSE operations, which makes the CPU the hottest). Run them overnight, and check that your PC is still running (and that all the finished work units sent didn't show any math error) the next morning...
You could also run a stress-test for the DDR SDRAM, using
MemTest86. Here again, run it overnight, and in UEFI mode (to get the latest tests, such as the row hammering one).