Session chair: Michael Gerndt, Tu Muenchen, Germany

Some Humans Generate Better Code Than Compilers or Machines: Leveraging Their Expertise
Prof. Nelson Amaral, Univ. of Alberta, Canada

Session chair: Prof. João Bispo, FEUP, Porto, Portugal

Towards an Efficient Unified Programming Model for Heterogeneous Computing
Pablo Antonio Martínez, José Manuel García and Gregorio Bernabé, University of Murcia, Murcia, Spain

Accelerating Compute-Intensive Serverless Functions
Mohak Chadha, Anshul Jindal and Michael Gerndt, TU Munchen, Germany

HOBFLOPS CNNs: Hardware Optimized Bitslice-Parallel Floating-Point Operations for Convolutional Neural Networks
James Garland and David Gregg, Trinity College, Ireland

Portability and efficiency in the runtime of the Controller heterogeneous programming model: The example of FPGAs
Arturo Gonzalez-Escribano, Yuri Torres de La Sierra, Francisco J. Andújar and Gabriel Rodriguez-Canal, University of Valladolid, Spain

• Prof. Alexandra Jimborean, University of Murcia, Spain
• Prof. Nelson Amaral, Univ. of Alberta, Canada
• Dr. Alex Zinenko, Google Inc, Paris, France

Chair: Prof. Paul Kelly, Imperial College of Science and Medicine, London, UK

The Role of Compiler for Parallel Computing in the Future

Over the last years there has been a push towards custom solutions and even high-profile applications with specific languages and frameworks (like TensorFlow). What do you think is the role of Compilers for these specific-domains, and what features should the language exhibit to allow compilers (and other productivity tools) to support predictable performance, correctness and security? Will we see the rise of a myriad of domain-specific languages and the corresponding compilers? In addition, there has been a wide application of ML and Data Mining techniques to software repositories begging the question: “Are we really going to need that many programmers with concurrent programming skills in the future”, and “What are the valued skills of the future programmers?” Most of us, in Academia, have struggled with how to keep up with the fast pace in this field, and are constantly faced with the challenges on how to best prepare our students on topics of compilation and translation for future computing technologies. What does the Panel think the disruptive models are or will be, beyond Quantum Computing, that will require us to think beyond existing languages?