In continued efforts to optimize Deep Learning workloads on Intel® architecture, our engineers explore various paths leading to the maximum performance. Not long ago, a technical preview of optimized AlexNet training on Caffe was published. Now we are sharing another preview of our work completely focused on the classification path and bringing it to new performance levels never demonstrated before on an Intel CPU.
The release of PARDISO* 4.0 from the University of Basel (UB) is not backward compatible with PARDISO from the Intel® MKL 10.3 and earlier and thus introduces some incompatibilities with the implementation of PARDISO that is available with the Intel® MKL
There are a few existing open source C++ template libraries that can be linked with Intel® MKL. Please refer to the documentation placed on the web-pages of the C++ libraries. Feel free to choose the package that mostly fits your needs and/or C++ style.
To obtain a sequence of random numbers with Intel® MKL VSL, you should initialize the Basic Random Number Generator (BRNG) that will be used in your application. Different types of VSL generators require a different number of seed values.
When linking application against PDE Poisson solver or DFTI solver you may obtain error messages like this: error LNK2019: unresolved external symbol dfti_commit_descriptor_external referenced in function MAIN__ error #7002: Error opening the module