Symposium on Energy Efficiency Challenges for HPC Systems 2019

One-Day Symposium on Energy Efficiency Challenges for HPC Systems

Date: 17 May 2019

We are pleased to invite you for the One Day Symposium on Energy Efficiency Challenges for HPC Systems on 17 May 2019 in Bengaluru.

PREAMBLE

The designing of High-Performance Computing (HPC) System has multidimensional challenges. Especially, Power and energy efficiency is one of the significant challenges in the burgeoning path of Exascale system. Excessive power consumption is a key limitation for further scaling of HPC Systems. Hence, an Adaptive Power Management Framework is essential for HPC System design and operation for better energy efficiency. Self-awareness, autonomous decision-making and the ability to characterize workloads online, the provision of pluggable control algorithms and provision of a fine-grained spatial and temporal power measurement empowered with seamless visualization and powerful analytics techniques are the essential aspects of an efficient power management system. Furthermore, Energy and Power-Aware Job Scheduling and Resource Management are also crucial for improving energy efficiency. In addition, Artificial Intelligence (AI) is going to be crucial in the next-generation HPC System operations and management. In particular, reinforcement learning (RL) based control algorithms can be significant for effective power management.

Achieving Exaflop performance within the 20-megawatt of targeted power consumption is a very daunting task. Moreover, the high variability in loads due to recurrent transition between phases of computation of HPC applications requires different power levels at a different timeline. Multi-MW step variations in loads would affect the entire ecosystem which includes power grid and utilities as well. In this perspective, the integration, coordination and orchestration of HPC facilities and the Power Grid have become crucial for reducing energy costs and carbon emissions. The Power Variability Prediction Models would be very important in this regard.

The key objective of this symposium is to provide a platform for researchers, academicians and industry to share their thoughts and to give their insights in building the next generation energy-efficient HPC Systems. C-DAC would like to take the viable approaches forward to build energy-efficient HPC System under the National Supercomputing Mission (NSM).

Focus Areas:

  • Energy Efficiency challenges for HPC
  • Adaptive Runtime System & Power Management Framework for HPC
  • Energy and Power-Aware Job Scheduling and Resource Management
  • Novel control algorithms for optimizing energy consumption for parallel workloads
  • Effective Visualization and powerful Analytics Techniques for HPC workload
  • Power Variability Prediction Models and Protection Mechanism for Grid Stability due to variability in loads

Click here for Workshop Brochure
(File Type: PDF, File Size: 704 KB, Date: 17/05/2019)

Click here for Workshop Agenda
(File Type: PDF, File Size: 231 KB, Date: 25/04/2019)

Click here for Registration

FOR ANY QUERIES/ASSISTANCE CONTACT:

Tel: (O) 080 - 25093456/ +91-9535399009
Email: ee-hpcw@cdac.in

SYMPOSIUM COORDINATOR:

SUMIT KUMAR SAURAV
Phone: 080 - 25093456/+91 -9535399009
E-mail: sumitk@cdac.in

SYMPOSIUM VENUE:

C-DAC Knowledge Park,
No. 1, Old Madras Road, Near Metro Station,
Baiyappanahalli, Bengaluru - 560 038
Karnataka
Route Map: http://goo.gl/s5msvj