BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260730T152637Z
LOCATION:
DTSTART;TZID=America/Los_Angeles:20260728T124700
DTEND;TZID=America/Los_Angeles:20260728T124800
UID:dac_DAC 2026_sess315_RESEARCH1449@linklings.com
SUMMARY:Maple: Efficient Data-Transfer-Aware Mapping Exploration for NDP-e
 nabled Edge LLM Inference
DESCRIPTION:Xuefeng Hua and Cong Li (Peking University), Xin Si (Southeast
  University), Qiang Wu (Houmo AI), and Guangyu Sun (Peking University)\n\n
 Near-DRAM Processing (NDP) has demonstrated great \n    potential for memo
 ry-bound \n    operators in edge-side large language model (LLM) inference
 .\n    Featuring a xPU-NDP heterogeneous system, \n    existing designs co
 ncentrate on optimizing the execution flow \n    within processing units, 
 \n    yet ignoring the bottleneck arising from xPU-NDP data transfer. \n  
   To tackle this challenge, we propose Maple, an efficient mapping explora
 tion framework.\n    Given a specific workload and NDP architecture, \n   
  Maple \n    adopts an address-mapping-based description method to constru
 ct a \n    comprehensive\n    search space that\n    encompasses\n    reso
 urce grouping, \n    tensor partitioning and transfer binding, thereby\n  
   facilitating joint optimization of computation and data transfer.\n    E
 xperiments show that Maple achieves an improvement of up to 3.34$\times$ i
 n performance \n    and 1.49$\times$ in energy compared with existing appr
 oaches on mainstream NDP architectures.\n\nTopics: Quantum\n\nTrack: DES6.
  Quantum Computing, Systems\n\nSession Chairs: Hokeun Kim (Arizona State U
 niversity); BaekGyu Kim (Daegu Gyeongbuk Institute of Science and Technolo
 gy); Xiang Chen (Peking University); Sangyeob Kim (Yonsei University); Rob
 ert M. Radway (University of Pennsylvania); Prashanth Mohan (University of
  Pittsburgh); Xiaoxuan Yang (University of Virginia, Stanford University);
  Ramtin Zand (University of South Carolina); Hadi Kamali (University of Ce
 ntral Florida); Jiafeng Xie (Villanova University); Prabuddha Chakraborty 
 (University of Maine); Junichiro Kadomoto (University of Tokyo); and Yosuk
 e Ueno (RIKEN)\n\n
END:VEVENT
END:VCALENDAR