2016 has been a busy initial year for High End Compute consultancy, and I would like to pass my thanks to each of you for your support. As I look forward to 2017, and the opportunities that it brings to work with several of you, I realise that I have capacity to do further work and would welcome discussing potential work or collaborations (full contact details in signature).
The key services that High End Compute provide are:
- Port & Support: helping your code run efficiently (in terms of performance and energy consumption) on your current system or on a new system, including advice on relative performances of various systems. Our service includes provision of a ported code/Make system but also hands-on passing over of the skills such that you can do likewise for other codes;
- Opt & Par: we can optimise and parallelise codes (FORTRAN and C) for various parallel paradigms (MPI, OpenMP, pThreads, OpenACC, OpenCL, CUDA), again with upskilling of yourselves;
- Training (& Outreach): using our decades of experience in designing and delivering training, both face to face as well as online & blended, we are uniquely positioned in the UK to provide bespoke courses that match your requirements
As an example of what we have achieved during the last 10 months:
- Design and delivery of outreach/training to ~30 schoolchildren at the Museum of Science & Industry to introduce parallel programming and leading them through each group constructing a parallel cluster comprising Raspberry Pi compute cards that they then used to program Python using MPI to scale a sieve of prime numbers. An event that has been picked up by GCHQ for further rollout nationwide;
- Chair of the Emerging Tech (EMiT) conference series that held its biggest and best international conference to date, at Barcelona Supercomputing Centre in June 2016, with Valencia lined up for 2017 and Jack Dongarra joining the Advisory Panel;
- Porting of an atmospheric chemistry code to the Knights Landing (KNL) node (Intel’s latest Xeon Phi) recently purchased by High End Compute (access available – drop us an email) to quantify energy savings compared to current compute platform;
- Design and delivery of an online training video for Intel, via one of our partnerships, with PSS, that outlined how to use OpenMP on their multicore chips and how to use the VTune Amplifier tools to identify and address load imbalance to enhance parallel performance;
- Design and delivery of a 2 day course for National Environment Research Council (NERC) that introduced their postdocs/grads, via innovative hands-on exercises, to the concepts of parallel computing, including OpenMP, OpenACC and MPI on the national Archer facility as well as using the Intel HPC Labs’ KNL node and NVIDIA’s labs.
Wishing you the best for the season and for 2017
Michael Bane, High End Compute
Contact Us For More Information