Cluster History

After several years of planning, Research Computing (originally known as the HPC Center) first became operational at UF in August 2004 with Phase I of the HPC cluster. Researchers quickly embraced high-performance computing so the cluster has been upgraded and expanded several times since to keep up with changing technology and user demand.

For historical information about the development of UF Research Computing, Annual Reports are available from 2006.

COMPUTE CLUSTER HISTORY

Phase	Date	Vendor	Mfgr Model	CPU Model	Core Count	Top 500 Score	HPL RMAX	HPL RPEAK
I	2004 Q4	Dell	PowerEdge 1750	Intel Xeon 2.8	400	225	1,325 GFlops	2,240 GFlops
IIa	2005 Q4	Rackable	C1001	AMD Opteron 275	832	419	3,000 GFlops	3,700 GFlops
IIb	2006 Q4	Rackable	C1001	AMD Opteron 275	480	N/A
III	2008 Q3	Penguin	Relion R1670SA	Intel Xeon E5462 2.8GHz	896	N/A
IV	2011 Q1	Dell	PowerEdge C6105	AMD Opteron 4184 2.8GHz	1,536	N/A
HPG1	2013 Q2	Dell	PowerEdge C6145	AMD Opteron 6378 2.4GHz	16,384	493	119,300 GFlops	157,300 GFlops
HPG2	2015 Q4	Dell	PowerEdge SOS6320	Intel Xeon E5-2698v3 2.3GHz	29,312	113	738,100 GFlops	1,074,000 GFlops
HPG3	2021 Q1	Lenovo	ThinkSystem SR645	AMD EPYC 7702 2.0GHz	30,720	N/A
HPG3	2021 Q2	Lenovo	ThinkSystem SR645	AMD EPYC 75F3 3.0GHz	9,600	N/A

GPU CLUSTER HISTORY

Phase	Date	Vendor	Mfgr Model	CPU Model	Core/Card Count	Top 500 Score	HPL RMAX	HPL RPEAK
HPG2	2015 Q4	Dell	PowerEdge R720	nVidia Tesla K80	40	N/A
HPG3	2019 Q2	Exxact	Tyan S7109	nVidia GeForce 2080TI	96	N/A
HPG3	2019 Q3	Exxact	Tyan S7109	nVidia GeForce 2080TI	160	N/A
HPG3	2019 Q4	Exxact	Tyan S7109	nVidia Quadro RTX 6000	48	N/A
HPG3	2019 Q4	Exxact	Tyan S7109	nVidia GeForce 2080TI	304	N/A
HPG-AI	2021 Q1	nVidia	DGX A100	nVidia Ampere A100 - 80GB	1,120	22	17,200 TFlops	21,314.7 TFlops

POSTERS

2007 poster advertising the HPC Center. This would have been Phase II.

OTHER PROJECTS

UF Apps For Research (2013-2023)

While UFApps provides a Virtual Desktop Interface (VDI) service for students and faculty to easily access and work with a large number of applications, it is not designed for faculty and graduate students to carry out complex calculations. To meet that need, UFIT Research Computing ran a separate version of UFApps where complex computations could be carried out on large data sets.

UF Apps for Research ran on Research Computing systems for ten years from December of 2013 until July of 2023.

Knights Landing (2016-2019)

Two Knights Landing (KNL) machines were acquired from Intel for a pair of fellowships concerning Intel Code Modernization. The purpose of this project was to optimize code for the Intel Xeon Phi. This was a one-year appointment starting in August of 2016. Code was developed on the systems until their retirement early in 2019.