Intel® MKL 11.0 introduces new functionality allowing users to balance the need for reproducibility in floating point results with performance. This webinar will explain the challenges of getting reproducible results in high performance floating point math libraries, introduce the controls in this new feature called Conditional Numerical Reproducibility (CNR) and discuss some of the performance results in the release version of the product.
Listen to a recording, download the materials, and review the Q&A.
If you have more questions, reply to this thread.
