v3.1.0

add cl_mem to kernel call. not sure if this is a good idea...

v2.2.1

Bug fix:

• doest produce display corruption on HD5500 now

v2.2.0

New:

• added devicedirtyflag to CLWrapper objects
  • starts as false
  • set to true anytime passed to kernel via out or inout, and that kernel is run
  • reset back to false by copyToHost() and copyToDevice() methods
  • so, can use to know when the host-side data is up to date or not

v2.1.0

New:

• added kernel store: storeKernel( string name, CLKernel *kernel ), getKernel( string name ), kernelExists( string name ), to facilitate per-connection kernel caching
• Added getCl() to CLWrapper types

Changes:

• changed from '-c++11x' to '-c++0x', in build, on linux, to allow build on more compilers

v2.0.0

Changes:

• renamed project from OpenCLHelper to EasyCL
• start semantic versioning

v1.6

• added getBuffer() to Wrapper objects, so can pass into clBLAS
• added a sample/test for using with clBLAS

v1.5

• fix bug where if both cpu and gpu available, it picks the cpu :-P
• fixed a crash bug related to platform_id initialization
• displays the platform and device chosen, at startup

v1.4

• generalized so can use any opencl-enabled device, and any platform
• builds on Windows (and linux ubuntu 14.04)
• unittests run on Windows now (and linux ubuntu 14.04)

Tested on:

• ubuntu 14.04, beignet
• ubuntu 14.04, nvidia grid k520
• windows 7, visual studio 2013, win32, intel cpu

v1.3

• builds on windows now
• some unittests run ok on windows too (but some dont :-P )

v1.1

v1.1