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:20260728T124600
DTEND;TZID=America/Los_Angeles:20260728T124700
UID:dac_DAC 2026_sess315_RESEARCH1692@linklings.com
SUMMARY:CORE: Compact, Optimized, and Resource-Efficient Compute Engine fo
 r FP-INT GEMM Acceleration with Binary Coding Quantization
DESCRIPTION:Chaewon Park, Inseong Hwang, Beom Jin Kang, and Hyun Kim (Seou
 l National University of Science and Technology)\n\nRecent advances in lar
 ge language models (LLMs) have driven exponential growth in parameter coun
 ts, amplifying memory footprints and creating acute computational bottlene
 cks. \nWeight-only quantization is a common mitigation because weights dom
 inate storage and incur less accuracy degradation than activation quantiza
 tion. However, it leaves activations in floating point (FP), forcing FP ex
 ecution whose area and energy costs far exceed those of integer (INT) comp
 ute. To address this limitation, binary coding quantization (BCQ) lowers t
 his cost by replacing many FP multiplies with FP additions, but still requ
 ires FP multiplications for scaling factors and suffers from limited repre
 sentational capacity, causing non-trivial accuracy loss. In this paper, we
  propose a sum-of-power-of-two scaling factor–based BCQ (SS-BCQ) that appr
 oximates FP scaling factors with a small set of power-of-two terms, elimin
 ating FP multiplications (shift–add only) while preserving accuracy. To ef
 ficiently realize SS-BCQ in hardware, we introduce CORE, a fully pipelined
  FP–INT general matrix-matrix multiplication (GEMM) engine. CORE features 
 (i) an adaptive data-mapping module that computes optimized block-wise act
 ivation scales to minimize scaling operations, and (ii) an extra processin
 g unit that allocates a small subset of processing elements to high-import
 ance weights to retain accuracy at low cost. Evaluated on the OPT-6.7B mod
 el, SS-BCQ reduces perplexity by 3.14 compared with prior BCQ methods. Imp
 lemented in CORE, our approach achieves up to 1.41x higher area efficiency
  (TOPS/mm^2) and 2.18x higher energy efficiency (TOPS/W) than state-of-the
 -art FP–INT accelerators, enabling a homogeneous low-precision execution p
 ath for on-device LLMs.\n\nTopics: Quantum\n\nTrack: DES6. Quantum Computi
 ng, Systems\n\nSession Chairs: Hokeun Kim (Arizona State University); Baek
 Gyu Kim (Daegu Gyeongbuk Institute of Science and Technology); Xiang Chen 
 (Peking University); Sangyeob Kim (Yonsei University); Robert M. Radway (U
 niversity of Pennsylvania); Prashanth Mohan (University of Pittsburgh); Xi
 aoxuan Yang (University of Virginia, Stanford University); Ramtin Zand (Un
 iversity of South Carolina); Hadi Kamali (University of Central Florida); 
 Jiafeng Xie (Villanova University); Prabuddha Chakraborty (University of M
 aine); Junichiro Kadomoto (University of Tokyo); and Yosuke Ueno (RIKEN)\n
 \n
END:VEVENT
END:VCALENDAR