High-Performance Medical Computing Core

The initial architecture of Geisinger's computing core includes 384 computational cores along 48 Intel Xeon central processing units, as well as 32 Nvidia Tesla general-purpose graphics processing units. The cluster is equipped with 2.2 terabytes of memory and a high-speed storage server with 248 terabytes of dedicated storage for investigational databases.

The high-performance core specializes in genomic analytics, machine learning, data visualization, predictive analytics, image analytics and advanced decisions support algorithms.

Geisinger has implemented cutting-edge enterprise data analytics frameworks, such has Hadoop 2 and the Berkeley Data Analytics Stack (BDAS), a major platform for tackling the next generation of problems coming out of Big-Data-style analyses and methodologies.

The infrastructure includes MPI, a high-performance library for message-passing between nodes; various distributed numerical libraries, such as PETSc, Intel MKL and Matlab MDCS; and experimental frameworks for tackling the next tier of exploratory data analysis, through interfaces such as Python and Julia. Moreover, Geisinger has the capability to exploit massively parallel computational hardware, such as GPGPUs and Intel Xeon Phi, and have personnel with experience in utilizing these methods at world-renown supercomputing centers, such as TACC Stampede and Kraken at Oak Ridge.

Core Co-Directors are Marylyn D. Ritchie, PhD, and John R Wallace, MS.

Join our team

The core is always looking for talented people to add to the team.

Send a resume and cover letter to jrwallace2@geisinger.edu.