Skip to content

Latest commit

 

History

History
39 lines (24 loc) · 1.19 KB

HISTORY.md

File metadata and controls

39 lines (24 loc) · 1.19 KB
Raw

What's New

v0.6.1 (branch master, benchmarking)

  • Adopt new syntax of GTScript for compile-time conditionals.
  • Update the annotations for the field arguments of the stencil definition functions- at run-time.

v0.6.0 (branch new_calling_api)

  • Migrate to the new stencil calling API of GT4Py.
  • Support to both device and managed memory for GPU storages.

v0.5.0 (branch new_storage_api)

  • Migrate to the new storage API of GT4Py.
  • GPU storages are allocated as managed memory. Because of some bugs in the implementation of the ufuncs, the isentropic model does not run with the gtcuda backend as the state cannot be initialized.

v0.4.0 (branch gt4py_v04_2)

  • Pass dedicated storages to the stencils. The storages do subclass numpy.ndarray.
  • Use Unified Memory Address (UMA) device memory.

v0.3.0 (branch gt4py_v04_1)

  • Provide stencils with numpy.ndarrays.
  • Avoid copying input fields into pre-allocated buffers.

v0.2.0 (branch gt4py_v04_0)

  • Upgrade to GT4Py-v0.4.0.
  • Pass dedicated storages to the stencils. The storages do not subclass numpy.ndarray.

v0.1.0 (branch gt4py_v03)

  • Leverage GT4Py-v0.3.0.