mkbane's blog

HEC Services: Many Faces of Consultancy

As well as hands on help getting models & data analyses to run faster or to scale out to larger problems, High End Compute can also provide expert opinion and help with writing research & funding grants, reducing your carbon footprint (and helping save money!), and strategic, independent, evaluation and requirements of your research/technical computing support portfolio.

To hear more, just drop us a line

HEC Services: Port ‘n’ Support

High End Compute doesn’t just want to take a code, move it to another architecture or make it faster, give it back to you, take a cheque and walk away. We are keen to work with you to help pass on skills so that in future you can do it yourself or at least understand much better what is involved. And when we provide training we focus on your examples and offer a follow-on to discuss if there are further options for you to improve your model or data analysis.

To find out more, contact us

Taking Stock of Our First Year

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:

Raspberry Pi Bramble - quick setup

Over the winter nights I've been working to set up a straightforward procedure for anybody to build a cluster out of Raspberry Pi boards, the PiHub, a few USB (power) leads, a cheap un-managed LAN box and a couple of LAN cables. By using MPI for Python via the Rasbian python-mpi4py package it is now within reach for us all to do both distributed computing (using MPI) between RPis as well as threaded (using OpenMP) on the 4 cores of the Pi 2.

How Green is Your Jetson?

Very much a work in progress

To determine how (power) efficient our new Jetson is for numerical computations, we shall consider a few things, one at a time:

  1. A single core of the ARM Cortex CPU
  2. The low power core of the Cortex
  3. The NVIDIA GPU
  4. All the CPU+GPU cores on the Jetson

As well as what to measure, there are questions over how to measure:

Benchmarking your high end compute

So, you've bought a new accelerator (maybe one of the dev XPhi on promo in Q4 2014?) or co-processor, or commissioned a cluster.

What's the first thing you want to do? Yes, benchmark every aspect to check it all works as you expect.

You would probably use the Intel MPI Benchmark suite to see how well your MPI implementation is working. There's also the HPCC Challenge benchmarks to determine how typical applications might run.

Subscribe to RSS - mkbane's blog