IceCube Institutional Memorandum Of Uderstanding (MOU)

Cowen is a co-PI on a successful $1M GPU-centric MRI proposal (“CyberLAMP”) to NSF along with colleagues in astronomy, materials science, and computer science. This MRI will provide considerable resources to IceCube computing. In anticipation grant approval, the handful of GPUs we purchased in April 2016 have enabled us to benchmark IceCube code on single- and double-precision GPUs (Titan X, K40 and K80) and will allow us to optimize the GPU purchase for the MRI accordingly. Cowen has also applied for and received additional local CPU resources in support of IceCube and PINGU related work. (As of September 2016, the quantity has not been specified, but it will be less than the requested 500 core-months.)

We will use these resources to contribute to simulation production, including simulation of low energy neutrinos with GENIE, PINGU simulations and reconstructions, and simprod jobs run collaboration-wide. Substantial amounts of reconstruction development work will also be conducted using these resources. It is not yet possible to specify the exact amounts of the contribution due to uncertainties in the final configuration of CyberLAMP and the unknown quantity of additional local resources we will be awarded.

Note: The activities and staffing levels in this MoU are appropriate for the period beginning October 1, 2016.

The numbers in the table below in parentheses are optimistic numbers that will obtain if applications for new computational resources are approved. The smaller number is the guaranteed average number of CPUs or GPUs, the second number after the dash is the possible burst level that we could obtain. The first number in parentheses is for the local resource application, the second for the NSF MRI (see text above for details). While the main emphasis will be on PINGU, GPU-centric IceCube jobs would certainly be welcome on the MRI cluster, if it is approved.

Labor Cat.	Names	WBS L3	Tasks	Funds Source	WBS 2.1	WBS 2.2	WBS 2.3	WBS 2.4	WBS 2.5	WBS 2.6	Grand Total
					Program Coordination	Detector Maintenance & Operations	Computing & Data Management	Data Processing & Simulation	Software	Calibration
KE	COWEN, DOUG	Education & Outreach	Education & Outreach	Inst. In-Kind	0.05						0.05
		Engineering and R&D	PINGU Co-Lead, Publication Committee	Inst. In-Kind	0.45						0.45
	COWEN, DOUG Total				0.40						0.50
SC	ANDERSON, TYLER	Data Acquisition	DAQ Firmware Development	NSF M&O Core		0.23					0.23
	ANDERSON, TYLER Total					0.23					0.23
PO	ELLER, PHILIPP	Simulation Production	Simulation Production	Base Grants				0.08			0.08
		Reconstruction	Develop analysis tools for systematics study; PISA development	Base Grants					0.20		0.20
		Detector Monitoring	Monitoring Shifts	Base Grants		0.03					0.03
		Central Computing Resources	Coordination and Support Grid distributed computing	NSF M&O Core			0.25				0.25
	ELLER, PHILIPP Total					0.03	0.25	0.08	0.20		0.56
	KEIVANI, AZADEH	Reconstruction	Maintain IceCube integration with AMON; HESE reco	Inst. In-Kind					0.25		0.25
	KEIVANI, AZADEH Total								0.25		0.25
	TESIC, GORDANA	Reconstruction	Maintain IceCube integration with AMON; HESE reco	Inst. In-Kind					0.25		0.25
	TESIC, GORDANA Total								0.25		0.25
GR	HUANG, FEIFEI	Reconstruction	Develop analysis tools for systematics study; PISA development	Inst. In-Kind					0.20		0.20
		Detector Monitoring	Monitoring Shifts	Base Grants		0.03					0.03
	HUANG, FEIFEI Total					0.03			0.20		0.23
	LANFRANCHI, JUSTIN	Reconstruction	Low energy event reconstruction quality; PISA development; Software Strike Team member	Inst. In-Kind					0.47		0.47
		Detector Monitoring	Monitoring Shifts	Inst. In-Kind		0.03					0.03
	LANFRANCHI, JUSTIN Total					0.03			0.47		0.50
	PANKOVA, DARIA	Data Acquisition	DAQ electronics hardware and firmware; background studies	Inst. In-Kind		0.47					0.47
		Detector Monitoring	Monitoring Shifts	Inst. In-Kind		0.03					0.03
	PANKOVA, DARIA Total					0.50					0.50
PSU Total					0.40	0.82	0.25	0.08	1.37		2.92

	2016		2017
	CPU Cores	GPU Cores	CPU Cores	GPU Cores
IceCube			?	?
PINGU	50-800 (40; 400)	1-6 K80, 1 TitanX (0; 50)	?	?
High E Array