How to extract block index and thread index of an error triggering kernel instance?


When I ran some kernel, it reports
“CUDA grid launch failed: CUcontext: 248763184 CUmodule: 142358743 Function: _Z28TestKernel9__StatesxPy4__Var2”

System Configuration: Win7, Visual Studio 2010, CUDA 6.5, Nsight 4.6, GTX Titan

I need to extract block and thread index that triggers the error so that I can do conditional kernel instance debugging to find out the faulty instruction. I checked into nsights.log file, but it didn’t have any details of the error. Where is the “error log” of Nsight debug output?