BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201803Z
LOCATION:263-264
DTSTART;TZID=America/Chicago:20251118T103000
DTEND;TZID=America/Chicago:20251118T105200
UID:submissions.supercomputing.org_SC25_sess184_pap304@linklings.com
SUMMARY:mLR: Scalable Laminography Reconstruction Based on Memoization
DESCRIPTION:Bin Ma (University of California, Merced); Victor Nikitin (Arg
 onne National Laboratory (ANL)); Xi Wang (University of California, Merced
 ); Tekin Bicer (Argonne National Laboratory (ANL)); and Dong Li (Universit
 y of California, Merced)\n\nADMM-FFT is an iterative method with high reco
 nstruction accuracy for laminography but suffers from excessive computatio
 n time and large memory consumption. We introduce mLR, which employs memoi
 zation to replace the time-consuming Fast Fourier Transform (FFT) operatio
 ns based on the unique observation that similar FFT operations appear in i
 terations of ADMM-FFT. We introduce a series of techniques to make the app
 lication of memoization to ADMM-FFT performance-beneficial and scalable. W
 e also introduce variable offloading to save CPU memory and scale ADMM-FFT
  across GPUs within and across nodes. Using mLR, we are able to scale ADMM
 -FFT on an input problem of $2K \times 2K \times 2K$, which is the largest
  input problem laminography reconstruction has ever worked on with the ADM
 M-FFT solution on limited memory; mLR brings 52.8\% performance improvemen
 t on average (up to 65.4\%), compared to the original ADMM-FFT.\n\nTag: Pe
 rformance Measurement, Modeling, & Tools\n\nRecording: Livestreamed, Recor
 ded\n\nRegistration Category: Technical Program Reg Pass\n\nSession Chair:
  Ian Karlin (NVIDIA Corporation)\n\n
END:VEVENT
END:VCALENDAR