New in 8.0: Systems Interfaces & Deployment
Version 8.0 introduces new features to load functions from shared libraries, giving a new way to incorporate external code into
Mathematica. It also adds support for GPU computing with new links to the CUDA and OpenCL environments. Version 8.0 also contains new tools for working with C code, including C code generation, symbolic representation of C code, and
Mathematica functions to drive standard C compilers for the platforms on which
Mathematica is available.
LibraryFunctionLoad — load a function from a shared library into
Mathematica
LibraryFunction — a function that calls into a shared library
LibraryFunctionInformation — information on a library function
LibraryFunctionUnload,
LibraryUnload — unload functions and libraries
FindLibrary,
$LibraryPath — find libraries resolving system-specific file extensions
CUDAInformation — list all CUDA device information
CUDAImageConvolve — convolve images with specified kernel
CUDAFunctionLoad — load a user-defined function to run on a GPU using CUDA
OpenCLInformation — list all OpenCL device information
OpenCLFunctionLoad — load a user-defined function to run on a GPU using OpenCL
Export — export a
Mathematica-compiled function into C code
CCodeGenerate — export a
Mathematica-compiled function into C code
ToCCodeString — convert a symbolic C expression into a string of C code
CreateExecutable — create an executable from C code
CreateLibrary — create a library from C code
Performance Improvements in Parallel Computing
Parallelize (modified) — performance improvements, automatic distribution of definitions
LaunchKernels (modified) — kernel-launching performance improvements
DistributeDefinitions (modified) — definition-distribution performance improvements
Standalone
Mathematica Scripts
-script — execute the commands from a file in a script mode
MathematicaScript — execute a standalone
Mathematica script
$ScriptCommandLine — arguments given to the
Mathematica script
