We have a high performance Bull Sequana X1000 supercomputer provided by Atos. The theoretical peak compute power of the system is 4.3 Petaflops (4.3 million billion operations per second!). That would take a human doing a calculation every second over 136 million years (roughly from the end of the Jurassic period to now).
The Spruce system supplied by SGI, is our latest supercomputer utilising SGIs M-Cell architecture but uses Intel Ivy-Bridge (x86-64) ten-core processors. The theoretical peak computer power of Spruce is 1.8 petaflops.
Visualisation, in the context of high performance supercomputing, involves the interpretation and understanding of results generated by large-scale simulation codes running on our supercomputing platforms.
Future HPC technologies
In the past, HPC technologies only delivered performance based on power-hungry architectural changes like clock frequency or increases in memory. More recently, the focus has changed to providing performance/watt. This is then linked to step gains for scientific applications and the insight they deliver towards new science, within a regime of energy efficiency, environmental balance and cost-effectiveness.
These balanced drivers are much more likely to deliver sustainable HPC performance improvements, and achieve exascale levels of compute performance within the next decade.