High performance computing systems are increasingly using co-processor systems with Intel and Nvidia seen as a key tag team for big data workloads, according to IDC.
In a study detailing high performance computing (HPC) sites, IDC looked at 905 systems. In 2011, IDC profiled 488 HPC systems. The two year jump largely highlights how 67 percent of HPC sites are now focused on big data workloads, said IDC.
The study from IDC corresponds with the latest top 500 supercomputer ranking.
Bottom line: Big data use cases are becoming more business critical and that's putting HPC systems on the map for various CXOs looking to boost revenue and cut risk.
IDC found the following:
- 76.9 percent of HPC systems were using co-processors and accelerators, up from 28.2 percent in 2011.
- Intel Xeon Phi and Nvidia GPUs are running for the lead in HPC systems.
- Co-processor systems are used primarily for exploratory uses.
- 30 percent of available computing cycles were devoted to big data work.
- Storage is the fastest growing technology at HPC sites.
- Cloud computing is increasingly used for HPC workloads with 23.5 percent of sites using public or private clouds to crunch data. That tally is up from 13.8 percent in 2011. The upshot with that finding is that cloud computing---helped along by companies such as Cycle Computing---are democratizing HPC systems.