== Changes between nvidia-healthmon v2.1 and v2.2 ==

- Added test for black-listed drivers.
- Improved bandwidth test.
- Improved healthmon robustness.
- Added memory test.
- Added user configurable temperature warning level
== Changes between nvidia-healthmon v2.2 and v2.3 ==

- Changed the debug logging flags to '-d' and '-debug'
- Added flags to specify an output file.  Reporting is redirected here.
- Remove P0 check from the bandwidth test
- Improved verbose output information in cases where a test was skipped
- Added new test for testing whether peer-to-peer is supported between
  two GPUs.  Test then runs bandwidth test between the GPUs.
- Added default configuration in the config.ini file for current Kepler GPUs
  These are best guesses that should be adjusted based on measurements taken
  on the local system
- Added a check to warn if the bandwidth test is being run on a NUMA system
- Modified tests to minimize communication with non-target GPUs
     - This should allow healthmon to provide more information when one
       GPU is unstable
- Removed support for C1060, M1060, S1070
- Added support for K10, K20, K20X, K20Xm, K20c, K20m, K20s
