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:20211208T164511Z
LOCATION:3016
DTSTART;TZID=America/Los_Angeles:20211207T133000
DTEND;TZID=America/Los_Angeles:20211207T150000
UID:dac_DAC 2021_sess126@linklings.com
SUMMARY:Speculations on Routes to Boost Power and Energy Efficiency
DESCRIPTION:Research Manuscript\n\nThis session describes four methods of 
 designing and managing digital systems to improve their power, energy, and
  thermal properties. It opens with a lightweight machine learning techniqu
 e to improve the boostability of multicore systems, considering power, per
 formance, and temperature. The next presentation describes control of runt
 ime dynamic concurrency throttling and frequency boosting parameters to im
 prove OpenMP energy-delay products. The third presentation describes a des
 ign space exploration technique using hybrid static/dynamic speculation; i
 t considers most-significant input bits, depth-of-carry histories, and loc
 al thread identifiers to improve accuracy. The session closes with an impe
 dance-driven design space exploration tool for board-level power networks.
 \n\nST2GPU: An Energy-Efficient GPU Design with Spatio-Temporal Shared-Thr
 ead Speculative Adders\n\nKandiah\n\nModern GPUs employ thousands of cores
 , yielding higher performance but also higher power consumption. To meet p
 erformance targets while staying within a reasonable power budget, designe
 rs have to make these execution cores increasingly more power efficient. O
 ne way to increase their power efficiency...\n\n---------------------\nSyn
 ergically Rebalancing Parallel Execution via DCT and Turbo Boosting\n\nLor
 enzon\n\nThe increasing use of cloud and HPC systems put more pressure on 
 the efficient utilization of hardware resources to keep costs low. Many dy
 namic concurrency throttling (DCT) techniques have successfully used to tu
 ne the number of executing threads to better balance a parallel applicatio
 n according t...\n\n---------------------\nSPROUT - Smart Power ROUting To
 ol for Board-Level Exploration and Prototyping\n\nBairamkulov\n\nThe board
 -level power network design process is governed by system-level parameters
  such as the number of layers and BGA. Evaluating the impact of these para
 meters on power-performance-cost (PPC) requires significant time. SPROUT, 
 an automated algorithm for prototyping printed circuit board (PCB) po...\n
 \n---------------------\nSmartBoost: Lightweight ML-Driven Boosting for Th
 ermally-Constrained Many-Core Processors\n\nRapp\n\nDynamic voltage and fr
 equency scaling (DVFS)-based boosting is indispensable for optimizing the 
 performance of a thermally-constrained many-core processor. This paper int
 roduces a novel boosting metric that for the first time integrates the app
 lication-dependent voltage/frequency (v/f) sensitivities...\n\n\nTopic: ED
 A\n\nKeyword: Cross-Layer Power Analysis and Low-Power Design\n\nVirtual P
 rogram: Presented In-Person
END:VEVENT
END:VCALENDAR
