Improving Performance for Small Size Problems
The overhead of calling an
function for small problem sizes can be significant when the functionhas a large number of parameters or internally checks parameter errors. To reduce the performance overhead for these small size problems, the
functions and directly call or inline special optimized small-matrix kernels that bypass error checking.For a list of functions supporting direct call, see
Limitations of the Direct Call.
Intel® oneAPI Math Kernel Library
Intel® oneAPI Math Kernel Library
direct call
feature works in conjunction with the compiler to preprocess the calling parameters to supported
Intel® oneAPI Math Kernel Library
To activate the feature, do the following:
- Compile your C or Fortran code with the preprocessor macro depending on whether a threaded or sequential mode ofis required by supplying the compiler option as explained below:Intel® oneAPI Math Kernel LibraryModeIntel® oneAPI Math Kernel LibraryMacroCompiler OptionThreadedMKL_DIRECT_CALL-DMKL_DIRECT_CALLSequentialMKL_DIRECT_CALL_SEQ-DMKL_DIRECT_CALL_SEQ
- For Fortran applications:
- Enable preprocessor by using the-fppoption for Intel® Fortran Compiler.
- Include theFortran include fileIntel® oneAPI Math Kernel Librarymkl_direct_call.fi.
Intel® oneAPI Math Kernel Library
dgemm
, with matrix ranks smaller than 50).
Product and Performance Information
|
---|
Performance varies by use, configuration and other factors. Learn more at
www.Intel.com/PerformanceIndex.
Notice revision #20201201
|