Ticket #16 (closed defect: invalid)

Opened 3 years ago

Last modified 3 years ago

CUDA: correct C version of function not emitted for Double types

Reported by: rmukhtar Owned by:
Priority: major Milestone: AutoMap, prototype
Component: CUDA backend Version: 0.7.1.0
Keywords: Cc:

Description (last modified by tmcdonell) (diff)

In the file "Data/Array/Accelerate/CUDA/CodeGen.hs" the patterns for "postfix" defined in the function for ccall does not correctly handle 'TypeDouble'.

Proposed fix: Add patterns for TypeDouble and TypeCDouble.

Change History

Changed 3 years ago by tmcdonell

  • description modified (diff)

Changed 3 years ago by tmcdonell

  • status changed from new to closed
  • resolution set to invalid

The 'ccall' helper function is for calling primitive C functions such as abs() and sin(), which default to type double, with float specific versions postfixed with an f as in absf() and sinf(). 'postfix' follows the same behaviour, so double types require no additional work.

Note: See TracTickets for help on using tickets.