Scientific Computing (SciComp), the high-performance computing shared resource of the Fred Hutchinson Cancer Research Center (FHCRC) respectfully requests funds to upgrade the current 232-node, high- performance computing (HPC) cluster created in 2004 and expanded in 2006, 2010, and 2013. The proposed upgrade involves retiring the 40 oldest nodes, then adding 158 new compute nodes and their supporting equipment to achieve a cluster capacity of 316 nodes. The new system offers a 50% increase in processing power and 200% increase in memory capacity. The expanded capacity will enable deep and efficient analysis of our research studies and accommodate the recent 50% growth in computing intensive research, much of which is not possible on the current cluster. The core user group for the new HPC cluster consists of at least 24 NIH-funded research groups at the Center numerous areas whose biomedical research aimed at eradicating cancer and other diseases is dependent on computationally intensive technical approaches as diverse as protein and virus evolution, sequencing, modeling outcomes of prostate cancers, studies of the human microbiome, biomarker evaluations, modeling of cancers, de-novo genome and transcriptome assembly, mRNA, miRNA, and structural variant detection, drug discovery, modeling of infectious agents and pandemics, computational modeling of protein interactions, detection of genetic susceptibility loci using genome-wide genotyping and sequencing studies and genome- wide gene-environment (GxE) studies. Several of our major researchers are currently experiencing substantial delays in accomplishing their work using our current system. Others have projects that cannot be done at all on the existing instrument. (see Research Projects section for details). SciComp has operated the current HPC cluster for 10 years and has a staff with a combined over 100 years of experience. The proposed new HPC cluster will be installed in available space in an FHCRC datacenter. The expanded cluster will address both immediate and future needs of our user community, supporting NIH-funded research at the FHCRC.
Showing the most recent 10 out of 159 publications
