0 Replies Latest reply on Jun 20, 2008 1:34 PM by dukeleto

    crash when using doubles

      float2 program shifted to double2 leads to runtime crash

      I have a working program based on float and float2 kernels, which I wish to run
      in double precision.
      I have replaced simply all float and float2 declarations by double and double2 declarations.
      The resulting .br compiles properly, and runs with BRT_RUNTIME=cpu.
      However, with BRT_RUNTIME=cal, I get the following message:

      invalid buffer, unable to write to file
      Assertion failure: calcontext.cpp (807) ...

      Is there something I am doing incorrectly, or is this a cal problem?
      Many thanks