v0.8.8

• Fix argument promotion for printf
• Fix error behavior on passing CPU arrays to the GPU

v0.8.7

• add device switching callback

v0.8.6

Fixes for Julia 0.7 and 1.0.

v0.8.5

• Improve compatibility with Julia 0.7-beta2
• Add more TBAA information

v0.8.4

Performance improvements for shared memory and loop nests
Detection of common pitfalls (base intrinsics, recursion)
Bugfixes

v0.8.3

Switch CI over to GitLab.

v0.8.2

• @device_code macro for dumping all forms of generated code to disk
• much better error messages
• fixes for Julia 0.7

v0.8.1

0.7 fixes and compiler improvements

v0.8.0

Feature release:

• support for closures and callable objects (#176)
• ignore the return value of kernel functions (#178)
• multi-GPU support with lazy initialization of the CUDA API (#175)

v0.6.3

Fixes for latest CUDAapi.
Use MicroLogging