Kees van Berkel



part-time full professor parallel computing


SAN (System Architecture and Networking)

Computer Science and Mathematics, Technical University Eindhoven

email: c.h.v.berkel@tue.nl

date: 2017 November 19


Drawing Drawing Drawing

Work experience

Industrial track:

  • 1980-2006: Philips Research, since 2000 as research fellow;
  • 2006-2008: NXP Research, as research fellow;
  • 2008-2013: ST-NXP Wireless, ST-Ericsson, as fellow;
  • 2013-2015: Ericsson, as fellow.

Academic track:

  • 1996- .. : TUe, as part-time full professor.

Summary of my research

1986 - 1996: asynchronous computing

  • pioneered handshake circuits, an architecture for asynchronous VLSI circuits, based on a formal semantics;
  • led the team that developed the Tangram language and silicon compiler, as well as applications (error decoders, decryption engines, microcontrollers, etc.); (Using these tools, Philips and NXP produced multiple billions of ICs for smart cards, e-passports, etc.)
  • defended my PhD thesis Handshake Circuits under supervision of Martin Rem;
  • (co-)supervised 3 PhD theses (Ad Peeters, Hans v. Gageldonk, Frank te Beest);
  • (co-)authored 35+ publications, the book Handshake Circuits (Cambridge University Press), and about 10 US patents.

1996 - 2006: embedded vector computing

  • pioneered an embedded vector processor architecture "$(scalar \;\|\; SIMD) \times VLIW$", offering 100 operations per clock cycle, at 100GOPS/Watt; (IC production exceeds 30M. Meanwhile, other companies (CEVA, Tensilica) adopted similar approaches);
  • contributed to multiple application studies for multi-standard smartphone modems (2G, 3G, 4G, WLAN);
  • (co-)supervised 2 PhD theses (Esko Dijk, Martijn van de Horst);
  • (co-)authored 15+ publications and about 10 US patents.

2006 - 2016: embedded multi-core computing

  • researched real-time multi-tasking on heterogeneous multi-core processors, with multi-standard wireless modems as applications ("2G-3G-4G software defined radio");
  • contributed to data-flow techniques and tools for modeling, performance analysis, resource management (including power), and buffer allocation; (Ericsson was evaluating these tools for production use until it discontinued their Modem business);
  • (co-)supervised 2 PhD theses (Orlando Moreira, Hrishikesh Salunkhe) + 1 in progress (Alok Lele);
  • (co-)authored 15+ publications and several US patents.

2016 - : high performance computing

  • attended various courses in radio astronomy and (astro)physics;
  • analyzed the compute workload of radio astronomy imaging for the SKA radio telescope MPSoC'15;
  • performed an apple-to-apple benchmarking of GPU versus FPGA, based on their intrinsic architecture features, using rooflines as tool, and 2D FFT as application; MPSoC'16;
  • developed the StaccatoLab dataflow-programming language and tool set MPSoC'17;
  • initiated the StaccatoLab research program.


Graduate course VLSI Programming (Very Large Scale Integrated Programming), course module 2IMN35, 5 ects. Next course: spring 2018 (Q4)


See my Google scholar profile