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:20230712T175212Z
LOCATION:3010\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20230711T142500
DTEND;TZID=America/Los_Angeles:20230711T144000
UID:dac_DAC 2023_sess156_RESEARCH1056@linklings.com
SUMMARY:HAIMA: A Hybrid SRAM and DRAM Accelerator-in-Memory Architecture f
 or Transformer
DESCRIPTION:Research Manuscript\n\nYan Ding, Chubo Liu, and Mingxing Duan 
 (Hunan University); Wanli Chang (University of York); Keqin Li (State Univ
 ersity of New York); and Kenli Li (Hunan University)\n\nResearchers have p
 aid great attention to embedding arithmetic logic units into the memory ar
 ray to reduce the overhead of data movement during Transformer inference. 
 However, existing work does not consider the heterogeneity of parallelism 
 and resource requirements among layers in Transformer. This not only incre
 ases the inference latency, but also causes the problem of low resource ut
 ilization. To address the above issue, we propose a hybrid accelerator-in-
 memory architecture named HAIMA, in which SRAM and DRAM cooperate to speed
  up the inference. Compared to Newton and TransPIM, our proposed hardware-
 software co-design architecture achieves 1x-1.5x speedup.\n\nTopic: Design
 \n\nKeyword: In-memory and Near-memory Computing Architectures, Applicatio
 ns and Systems\n\nSession Chairs: Laura Fick (Mythic) and Biresh Kumar Joa
 rdar (Duke University)
END:VEVENT
END:VCALENDAR
