Move scikit-cuda notebook to its own directory

Author: gjbex

source-code/README.md

@@ -8,7 +8,8 @@ Sample code for performing computations on a GPU.
 1. `pycuda`: directory illustrating pyCUDA.
 1. `curand.ipynb`: Jupyter notebook illustrating generating random
    numbers on a GPU.
-1. `scikit_cuda.ipynb`: Jupyter notebook illustrating linear algebra
-   on a GPU device.
+1. `scikit_cuda`: code illustrating linear algebra on a GPU device.  This
+   is an older library and has been at least partially superceded by
+   CuPy.
 1. `numba`: code illustrating using numba for GPU computing.
 1. `cupy`: Jupyter notebook illustrating some aspects of the `cupy` package.

source-code/scikit_cuda/README.md

@@ -0,0 +1,15 @@
+# scikit-cuda
+
+Scikit-cuda provides a Python interface to many linear algebra routines on the
+GPU using CUDA. The interface is similar to that of
+[scikit-learn](http://scikit-learn.org/stable/), and aims to provide simple and
+efficient access to GPU operations for data science and scientific computing.
+
+However, the library is aging, and has at least partially been superceded by
+[CuPy](https://cupy.chainer.org/), which provides a more general-purpose
+interface to the CUDA API.
+
+
+## What is it?
+
+1. `scikit_cuda.ipynb`: Jupyter notebook with examples of using scikit-cuda.