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:20260402T024534Z
LOCATION:Level 2 Lobby
DTSTART;TZID=America/Los_Angeles:20250623T180000
DTEND;TZID=America/Los_Angeles:20250623T190000
UID:dac_DAC 2025_sess287_LBR111@linklings.com
SUMMARY:Late Breaking Results: Source-Aware Adaptive Cache Management for 
 CXL-enabled Disaggregated Memory Sharing
DESCRIPTION:Qianyu Cheng, JiaJun Ji, Teng Wang, Zihan Wang, Lei Gong, Chao
  Wang, and Xuehai Zhou (University of Science and Technology of China)\n\n
 Dynamic workloads running on multiple hosts will bring changing access pat
 terns on CXL-enabled shared disaggregated memory. Existing works often un-
 traceably cache multi-source accesses, making it hard to exploit each host
 's access behavior and assure service quality. Our solution Alchemy jointl
 y optimizes cache replacement and bypassing and runs as an online reinforc
 ement learning agent with source-aware adaptivity. It gives rewards derive
 d from sampling-based action effectiveness and per-host macro performance.
  The multi-host prototype-based results on FPGAs show 8.71%-14.56% reducti
 on in average access latency over LRU policy and 44x faster than the hardw
 are-efficient ICGMM method in decision-making with comparable overhead.\n\
 n
END:VEVENT
END:VCALENDAR