Thursday, 24 September 2015

Collective Knowledge is now available via PyPi

Dear colleagues,
 
Just a note that we finally added Collective Knowledge framework to PyPi archive. It should now be possible to install CK simply via:
 
$ pip install ck
 
The idea is to really minimize set up time for researchers before prototyping their ideas and sharing experimental results. For example, on Linux with GCC, it should take just 3 steps to compile and run a given shared benchmark:
 
$ ck pull repo:ctuning-benchmarks
$ ck compile program:cbench-automotive-susan --speed
$ ck run program:cbench-automotive-susan
 
Similarly, it is possible to run various shared scenarios (including compilation on Windows, cross-compilation forAndroid, multi-objective autotuning, predictive analytics, compiler bug detection) as described here:
Feedback and suggestions are very welcome! Also, do not hesitate to open enhancement or bug requests here:
 
Have fun,
dividiti team

1 comment:

  1. We just fixed a bug occuring when making fresh CK installation via PIP on some machines. You can get the latest V1.6.5 version via:
    $ (sudo) pip install ck -U

    If you use python3 and python2 at the same time on Linux, and you installed CK for python3, you may need to set CK_PYTHON to python3, i.e.
    $ export CK_PYTHON=python3

    ReplyDelete