Simple, but effective workaround to solve the 100% CPU burning loop bug in Nvidias OpenCL runtime, works Linux and Windows, 0% cpu (alsmost), 100% performance in comparison to CPU burning loop variant. Replaces old previous try with overwritten clock_gettime() function

2016-06-13 23:50:49 +02:00
parent 217f9e6004
commit 8abc42e346
2 changed files with 8 additions and 17 deletions
--- a/include/types.h
+++ b/include/types.h
@@ -988,6 +988,7 @@ struct __hc_device_param

  uint    exec_pos;
  double  exec_ms[EXEC_CACHE];
+  double  exec_prev;

  // this is "current" speed