BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20181221T160729Z
LOCATION:C141/143/149
DTSTART;TZID=America/Chicago:20181113T153000
DTEND;TZID=America/Chicago:20181113T160000
UID:submissions.supercomputing.org_SC18_sess212_pap511@linklings.com
SUMMARY:HiCOO: Hierarchical Storage of Sparse Tensors
DESCRIPTION:Paper\nAlgorithms, Graph Algorithms, Linear Algebra, Machine L
 earning, Sparse Computation, Tech Program Reg Pass, BSP Finalist\n\nHiCOO:
  Hierarchical Storage of Sparse Tensors\n\nLi, Sun, Vuduc\n\nThis paper pr
 oposes a new storage format for sparse tensors, called Hierarchical COOrdi
 nate (HiCOO; pronounced: “haiku”). It derives from coordinate (COO) format
 , arguably the de facto standard for general sparse tensor storage. HiCOO 
 improves upon COO by compressing the indices in units of sparse tensor blo
 cks, with the goals of preserving the “mode-agnostic” simplicity of COO wh
 ile reducing the bytes needed to represent the tensor and promoting data l
 ocality. We evaluate HiCOO by implementing a single-node, multicore-parall
 el version of the matricized tensor-times-Khatri-Rao product (MTTKRP) oper
 ation, which is the most expensive computational core in the widely used C
 ANDECOMP/PARAFAC decomposition(CPD) algorithm. This MTTKRP implementation 
 achieves up to 23.0× (6.8× on average) speedup over COO format and up to 1
 5.6× (3.1× on average) speedup over another state-of-the-art format, compr
 essed sparse fiber (CSF), by using less or comparable storage of them. Whe
 n used within CPD, we also observe speedups against COO- and CSF-based imp
 lementations.
URL:https://sc18.supercomputing.org/presentation/?id=pap511&sess=sess212
END:VEVENT
END:VCALENDAR

