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:20260402T024533Z
LOCATION:3000\, Level 3
DTSTART;TZID=America/Los_Angeles:20250624T134500
DTEND;TZID=America/Los_Angeles:20250624T140000
UID:dac_DAC 2025_sess101_RESEARCH1876@linklings.com
SUMMARY:APSQ: Additive Partial Sum Quantization with Algorithm-Hardware Co
 -Design
DESCRIPTION:Yonghao Tan, Pingcheng Dong, Yongkun Wu, Yu Liu, Xuejiao Liu, 
 Peng Luo, Shih-Yang Liu, Xijie Huang, Dong Zhang, Luhong Liang, and Kwang-
 Ting Cheng (The Hong Kong University of Science and Technology)\n\nDNN acc
 elerators, significantly advanced by model compression and specialized dat
 aflow techniques, have marked considerable progress. However, the frequent
  access of high-precision partial sums (PSUMs) leads to excessive memory d
 emands in architectures utilizing input/weight stationary dataflows. Tradi
 tional compression strategies have typically overlooked PSUM quantization,
  which may account for 69% of power consumption. This study introduces a n
 ovel Additive Partial Sum Quantization (APSQ) method, seamlessly integrati
 ng PSUM accumulation into the quantization framework. A grouping strategy 
 that combines APSQ with PSUM quantization enhanced by a reconfigurable arc
 hitecture is further proposed. The APSQ performs nearly lossless on NLP an
 d CV tasks across BERT, Segformer, and EfficientViT models while compressi
 ng PSUMs to INT8. This leads to a notable reduction in energy costs by 28-
 87%. Extended experiments on LLaMA2-7B demonstrate the potential of APSQ f
 or large language models. The codes will be open-source upon publication.\
 n\nTopics: AI\n\nTracks: AI1: AI/ML Algorithms\n\nSession Chairs: Fan Yang
  (Fudan University) and Li Shang (Fudan University)\n\n
END:VEVENT
END:VCALENDAR
