Dear R users,
I am happy to announce the most recent version of gpuR has been
released. There are several new enhancements to the package including
the ability to use user written OpenCL kernels. A full list of
changes from the NEWS are shown below.
API Changes:
1. deviceType, gpuInfo, cpuInfo not longer accepts 'platform_idx'
parameter as OpenCL contexts cannot contain more than one platform.
New Features:
1. Added functionality to create custom OpenCL functions from user
provided kernels
2. Added 'synchronize' function to assure completion of device calls
(necessary for benchmarking)
3. Added determinant function (det)
4. Allow for gpuR object - base object interaction (e.g. vclMatrix * matrix)
5. Added ?inplace' function for ?inplace' operations. These operations
include '+', '-', '*', '/', 'sin',
'asin', 'sinh', 'cos', 'acos',
'cosh', 'tan', 'atan', 'tanh'.
6. Added 'sqrt', 'sum', 'sign','pmin', and
'pmax' functions
7. Methods to pass two gpuR matrix objects to 'cov'
8. Added 'norm' method
9. Added gpuRmatrix/gpuRvector Arith '+','-' methods
Bug Fixes:
1. Fixed incorrect device info when using different contexts
2. Fixed Integer Matrix Multiplication
3. All OpenCL devices will be initialized on startup (previous version
occasionally would omit some devices)
There are many more features in the works. Suggestions and
contributions continue to be welcomed. Please submit all through my
github issues https://github.com/cdeterman/gpuR/issues
<https://github.com/cdeterman/gpuR.git>
Also, thanks to all those as well for testing this package on various
GPU devices and operating systems. A lot of the stability of this
package is made possible by your efforts.
Kind regards,
[[alternative HTML version deleted]]
