Predicting The Convergence Of HPC & AI In A Multi-Architecture Future

Software development for High Performance Computing and Artificial Intelligence in the data center needs already today to target multiple architectures, including CPUs, GPUs, FPGA and specialized ASICs processors. Depending on the architecture, software developers need to use specific languages, libraries, frameworks or APIs, which creates complexity and increases the costs of every project. In the future, with the convergence between HPC and AI as well as the increasing velocity of code development, the complexity will need to be dramatically reduced. Software innovation requires a new paradigm for this multi-architecture future!