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:20211208T164509Z
LOCATION:3018
DTSTART;TZID=America/Los_Angeles:20211207T133000
DTEND;TZID=America/Los_Angeles:20211207T143500
UID:dac_DAC 2021_sess116@linklings.com
SUMMARY:Smarter Compute Engines for Greener ML
DESCRIPTION:Research Manuscript\n\nThis world is always seeking better har
 dware accelerators to improve machine learning computing performance. This
  session presents four research works focusing on optimizing computer hard
 ware to improve machine learning execution performance. Efficient hardware
  accelerators incorporated with innovative optimizations are presented in 
 this session to accelerate widely-used machine learning networks, such as 
 2D and 3D convolutional neural networks as well as reinforcement learning,
  with less energy consumption.\n\nDataflow Mirroring: Architectural Suppor
 t for Highly Efficient Fine-Grained Spatial Multitasking on Systolic-Array
  NPUs\n\nChoi\n\nWe present dataflow mirroring, architectural support for 
 low-overhead fine-grained systolic array allocation which overcomes the li
 mitations of prior coarse-grained spatial-multitasking Neural Processing U
 nit (NPU) architectures. The key idea of dataflow mirroring is to reverse 
 the dataflows of co-l...\n\n---------------------\nRASA: Efficient Registe
 r-Aware Systolic Array Matrix Engine for CPU\n\nJeong\n\nAs AI-based appli
 cations become pervasive, CPU vendors are starting to incorporate matrix e
 ngines within the datapath to boost efficiency. Systolic arrays have been 
 the premier architectural choice as matrix engines in offload accelerators
 . However, we demonstrate that incorporating them inside CPUs...\n\n------
 ---------------\nFIXAR: A Fixed-Point Deep Reinforcement Learning Platform
  with Quantization-Aware Training and Adaptive Parallelism\n\nYang\n\nIn t
 his paper, we present a deep reinforcement learning platform named FIXAR w
 hich employs fixed-point data types and arithmetic units for the first tim
 e using a SW/HW codesign approach. Starting from 32-bit fixed-point data, 
 Quantization-Aware Training (QAT) reduces its data precision based on the 
 ...\n\n\nTopic: Design\n\nKeyword: AI/ML System Design\n\nVirtual Program:
  Presented In-Person
END:VEVENT
END:VCALENDAR