Date Wed 12 November 2008 Tags pycuda

I’m happy to announce the availability of PyCuda 0.91. There is full, up-to-date documentation available.

The following exciting stuff is in PyCuda 0.91:

  • Support for Windows and MacOS X, in addition to Linux. (Gert Wohlgemuth, Cosmin Stejerean, Znah on the Nvidia forums, and David Gadling)
  • Support more arithmetic operators on pycuda.gpuarray.GPUArray. (Gert Wohlgemuth)
  • Add pycuda.gpuarray.arange(). (Gert Wohlgemuth)
  • Add pycuda.curandom. (Gert Wohlgemuth)
  • Add pycuda.cumath. (Gert Wohlgemuth)
  • Add pycuda.autoinit.
  • Add pycuda.tools.
  • Add pycuda.tools.DeviceData and pycuda.tools.OccupancyRecord. pycuda.gpuarray.
  • GPUArray parallelizes properly on GTX200-generation devices.
  • Add support for compiling on CUDA 1.1. Added version query pycuda.driver.get_version(). Updated documentation to show 2.0-only functionality.
  • Make pycuda.driver.Function resource usage available to the program. (See, e.g. pycuda.driver.Function.registers.) Cache kernels compiled by pycuda.driver.SourceModule.
  • Allow for faster, prepared kernel invocation. See pycuda.driver.Function.prepare().
  • Added memory pools, at pycuda.tools.DeviceMemoryPool as experimental, undocumented functionality. For some workloads, this can cure the slowness of pycuda.driver.mem_alloc().
  • Fix the memset family of functions.
  • Improve Error Reporting.

Check the docs change list for a fully hyperlinked version of the above.

Have fun, Andreas