News

  • June 2022: We ran the 2nd ASTRA-sim tutorial at ISCA 2022.

  • May 2022: Sheng-Chun (Felix) successfully defended his PhD thesis titled “Domain-aware Genetic Algorithms for Hardware and Mapping Optimization for Efficient DNN Acceleration” becoming the fifth PhD alum from the group. Congratulations Dr. Kao – best wishes for the next phase of your career!

  • April 2022: Our lab’s first SIGMETRICS paper! Our formal metric called flexion for quantifying flexibility of DNN Accelerators got accepted at SIGMETRICS 2022. Congrats Felix!! (This work was in collaboration with Hyoukjun, Micheal Pellauer (NVIDIA) and Angshuman Parashar (NVIDIA). It extends our CAL 2021 paper on this topic.

  • April 2022: Dr. Krishna was inducted into the HPCA Hall of Fame! This distinction recognizes authors with eight or more papers that have appeared in HPCA till date.

  • March 2022: Our paper introducing “Self Adaptive Reconfigurable Arrays (SARA)” was accepted at DAC 2022! Congrats Anand!

  • March 2022: Our paper “Themis: A Network Bandwidth-Aware Collective Scheduling Policy for Distributed Training of DL Models” was accepted at ISCA 2022! Congrats Saeed and William! This was in collaboration with Sudarshan Srinivasan (Intel) and Srinivas Sridharan (Meta)
  • February 2022: We ran two tutorials at ASPLOS 2022. Check out the websites for more detail.
  • February 2022: Eric successfully defended his PhD thesis titled “Building Efficient Tensor Accelerators for Sparse and Irregular Workloads” becoming the fourth PhD alum from the group. Congratulations Dr. Qin and wish you lots of success in future!
  • January 2022: Our paper analyzing the design-space of dataflows for Graph Neural Networks on spatial accelerators got accepted at IPDPS 2022 and is a best paper candidate! Congrats to Raveesh, Eric and our collaborators from UPC, Univ of Murcia and Sandia National Labs!
  • November 2021: Our paper on DiGamma, a framework for co-optimizing DNN accelerator mappings and hardware got accepted at DATE 2022. Congrats Sheng-Chun (Felix)!
  • October 2021: Anand successfully defended his PhD thesis titled “Methodology and Analysis for Efficient Custom Architecture Design using Machine Learning” becoming the third PhD alum from the group. Congratulations Dr. Samajdar and wish you tremendous success in future endeavors!
  • October 2021: Two papers accepted at HPCA 2022! The first is an optimization framework called MAGMA for mapping multiple DNNs on multi-accelerator platforms – Congrats Felix! The second is a novel deadlock-freedom technique for NoCs – Congrats to an amazing collaboration team: Hossein, Natalie  (Toronto), Paul (TAMU) and Joshua (Wisc)!
  • October 2021: We co-organized the First Workshop on “Architecture, Compiler, and System Support for Multi-model DNN Workloads” in collaboration with Facebook Reality Labs at MICRO 2021. Detailed program and recorded video can be found here.

  • September 2021: Dr. Krishna won a Qualcomm Faculty Award!

  • September 2021: Our paper on STONNE, a cycle-accurate simulator for modeling flexible dense and sparse DNN accelerators got accepted at IISWC 2021! And a novel spatio-temporal reduction fabric implemented in STONNE got accepted at NOCS 2021 and is a best paper nominee! Congrats to our lab’s visiting student Francisco, and his advisors Prof. Jose L. Abellan and Manuel Acacio.
  • July 2021: Our paper on UNION, a framework in MLIR that unifies multiple accelerator cost models (MAESTRO and Timeloop) and mappers under a common framework accepted at PACT 2021! Congratulations to Geonhwa for leading the integration effort and our close collaborators: Prasanth (now at IBM), Angshu and Po-An (NVIDIA), Gokcen and Roberto (PNNL) and Siva (Sandia National Labs)
  • June 2021: Our paper on SEEC, a unified scheme for deadlock-freedom and throughput enhancement for NoCs, got accepted at SC 2021. It is also a best paper award nominee! Congrats to our lab’s recent alum Dr. Mayank Parasar!

  • June 2021: We ran a tutorial on SCALE-sim at ISCA 2021.

  • June 2021: Our lab’s recent alumnus Dr. Hyoukjun Kwon received an Honorable Mention at the 2021 ACM-SIGARCH / IEEE-CS TCCA Outstanding Dissertation Award ceremony at ISCA 2021! Congrats Hyoukjun!! We are really proud of your work.

  • Apr 2021: Dr. Krishna received tenure!!

  • Apr 2021: We ran a tutorial on SCALE-sim at ASPLOS 2021.

  • Apr 2021: Dr. Krishna co-organized a workshop “SysML4Health: Scalable Systems for ML-driven Analytics in Healthcare” at MLSys 2021 with Professors Alexey Tumanov, Vivek Sarkar (GT), Jimeng Sun (UIUC) and Dawn Song (UC Berkeley)

  • Mar 2021: Our paper titled “Enabling Compute-Communication Overlap in Distributed Training Platforms” got accepted at ISCA 2021! Congrats Saeed! Special thanks to our close collaborators from Facebook and Intel for introducing us to this research direction.

  • Feb 2021: Our paper on RASA – an efficient tensor core inside CPUs got accepted at DAC 2021! Congrats Geonhwa, Anand and Eric, and our collaborators Hyesoon Kim (Georgia Tech), Chris Hughes and Sree Subramoney (Intel Labs).

  • Jan 2021: Our paper titled “Extending Sparse Tensor Accelerators to Support Multiple Compression Formats” has been accepted at IPDPS 2021! Congrats Eric Qin,  student co-authors (Geonhwa, William, Felix, Hyoukjun), Intel co-authors (Sudarshan and Dipankar) and Sandia National Lab co-authors (Gordon Moon and Siva Rajamanickam)!

  • Nov 2020: Two papers accepted at HPCA 2021! Congrats to Hyoukjun for Herald and Hossein (from Univ of Toronto) for Pitstop!

  • Oct 2020: We ran a tutorial on MAESTRO at MICRO 2020 which had over 80 attendees! The tutorial website provides a link to all slides and videos.
  • Aug 2020: Dr Krishna, Hyoukjun Kwon, Michael Pellauer (NVIDIA), Angshuman Parashar (NVIDIA) and Ananda Samajdar co-authored a Synthesis Lectures on Computer Architecture titled “Data Orchestration in Deep Learning Accelerators” which was released by Morgan & Claypool publishers.
  • Aug 2020: Dr. Krishna gave an invited talk titled “Benchmarking DNN Models on Future Inference and Training Platforms“ at the MLBench workshop at ISPASS 2020.

  • Jul 2020: Dr. Krishna won a Facebook Faculty Research Award for AI System HW/SW Co-Design!

  • Jul 2020: Our paper on GAMMA – a genetic algorithm based framework for finding efficient mappings of DNNs on accelerators accepted at ICCAD 2020! Congrats Sheng-Chun (Felix) – on a roll!
  • Jul 2020: Our paper on ConfuciuX – a Reinforcement Learning method for Accelerator Design-Space Exploration got accepted at MICRO 2020! Congrats to Sheng-Chun (Felix) for his first paper!
  • Jul 2020: Hyoukjun successfully defended his PhD thesis titled “Data- and Communication-centric Approaches to Model and Design Flexible Deep Neural Network Accelerators” becoming the second PhD alum from the group. Congratulations Dr. Kwon and wish you a lot of success in future!
  • Jul 2020: Mayank successfully defended his PhD thesis titled “Subactive Techniques for Guaranteeing Routing and Protocol Deadlock Freedom in Interconnection Networks” becoming the first PhD alum from the group. Congratulations Dr. Parasar and best wishes for all your future endeavors!
  • May 2020: Dr. Krishna gave an invited talk titled “A Communication-Centric Approach for Designing Flexible DNN Accelerators“ at the AccML workshop at ISCA 2020.
  • Feb 2020: SIGMA won Best Paper Award at HPCA 2020!! Congrats Eric, Anand, Hyoukjun and Vineet!! More details here.
  • Feb 2020: Two papers accepted at DAC 2020! Congrats to Srikant and Hyoukjun!
  • Jan 2020: Three papers accepted at ISPASS 2020!! Congrats Anand and Moritz for SCALE-sim, Parth and Anand for CLAN, and Saeed for ASTRA-sim!
  • Jan 2020: MAESTRO (MICRO 2019) got selected as an IEEE Micro Top Picks! Congrats Hyoukjun and Prasanth!
  • Nov 2019: Three papers accepted at HPCA 2020!! Congrats Mayank for DRAIN, Eric, Anand, Hyoukjun and Vineet for SIGMA, and Bahar for ALRESCHA!
  • Nov 2019: Dr. Krishna selected to jointly lead $5.5 Million DoE Center on AI HW-SW Co-Design in partnership with Pacific Northwestern National Labs and Sandia National Labs.
  • Sept 2019: Dr. Krishna won Best Presenter at the ARM Research Summit 2019.
  • Sept 2019: Dr. Krishna appointed to the ON Semiconductor Professorship in the School of ECE at Georgia Tech.
  • Jul 2019: Two papers got accepted at MICRO 2019. Congrats Hyoukjun and Prasanth for MAESTRO and Mayank for SWAP!
  • Jul 2019: Paper on BINDU got into NOCS 2019. Congrats Mayank!
  • Jul 2019: We ran a Tutorial on MAESTRO at ICONS 2019
  • Jun 2019: We ran a Tutorial on SCALE-Sim at ISCA 2019.
  • May  2019: Our paper on interconnect-aware mapping of ML accelerators on FPGAs got accepted at FPL 2019!
  • May 2019: Dr. Krishna won a Facebook Faculty Award for AI System HW-SW Co-Design!
  • April 2019: We ran a workshop on Unlocking the Power of Edge Computing at ASPLOS 2019.
  • March 2019: Dr. Krishna won a Google Faculty Award!
  • February 2019: We ran a full-day tutorial on MAERI and MAESTRO at HPCA 2019.
  • January 2019: SPIN (ISCA 2018) got selected as an IEEE Micro Top Picks, while MAERI (ASPLOS 2018) got selected as an Honorable Mention! Congrats Aniruddh, Hyoukjun, Anand. The first set of papers from Synergy Lab to receive this honor.

  • January 2019: mRNA (a tool for automatically mapping DNNs  on MAERI got accepted at ISPASS 2019! Congrats Zhongyuan, Hyoukjun and Sachit on publishing our group’s first peer-reviewed conference paper for 2019.

  • October 2018: Dr. Krishna gave an invited seminar at the Oakridge National Laboratories (ORNL).

  • October 2018: Dr. Krishna co-organized the inaugural Rising Stars in Computer Architecture workshop at Georgia Tech.

  • October 2018: An extended version of MAERI got accepted in the IEEE Micro Special Issue on Hardware Acceleration, Nov/Dec 2018! Congrats to Hyoukjun and Anand.
  • October 2018: Dr. Krishna has won the “2018 Class of 1940 Course Survey Teaching Effectiveness Award” from Georgia Institute of Technology.
  • September 2018: Should DNN Accelerators be SIMD or Systolic? Check out Dr. Krishna’s SIGARCH blog article.
  • August 2018: Congrats to Mayank and Hyoukjun for getting selected to present at the ACM Student Research Competition (SRC) at MICRO 2018!
  • July 2018: Two papers accepted at MICRO 2018! Congrats Anand and Parth for GeneSys, and Srikant for NOCSTAR.
  • July 2018: Paper on Brownian Bubble Router for deadlock-freedom accepted in NOCS 2018. Congrats Mayank!
  • June 2018: Dr. Krishna gave an invited seminar at IBM T. J. Watson Research Center.
  • June 2018: Dr. Krishna gave a talk on MAERI at the Machine Learning in Science and Engineering  conference at CMU.
  • June 2018: We ran a tutorial on the MAESTRO analytic model and MAERI DNN accelerator at ISCA 2018.
  • May 2018: Dr. Krishna gave a talk on “Communication-centric Computing for Deep Learning” at the 2nd  Technical Exchange Conference at the Institute for  Electronics and Nanotechnology (IEN) at Georgia Tech.

  • March 2018: Three papers accepted at ISCA 2018! Congrats Mayank and Abhishek for SEESAW; Aniruddh and Paul for SPIN; and Nachiket for FastTrack!

  • March 2018: MAERI was presented in ASPLOS 2018. Slides and the poster are available for download.

  • February 2018: Hyoukjun and Anand will be doing demos on MAERI and SCALE-Sim respectively at Cogarch 2018.

  • February 2018: Paper on neuro-evolutionary accelerator accepted at Cogarch 2018! Congrats Anand!

  • January 2018: Dr. Krishna won the NSF CRII award!

  • January 2018: Paper on the characterization of the NoC within the Micron Hybrid Memory Cube (HMC) accepted to ISPASS 2018! Congrats Ramyad!
  • January 2018: MAERI got accepted at SysML 2018! Congrats Hyoukjun and Anand!
  • November 2017: Two papers accepted at ASPLOS 2018! Congrats Hyoukjun and Ananda for MAERI; and Mohan, Steffen and Sanidhya for LATR!
  • November 2017: Hyoukjun gave a Demo on OpenSMART at the OpenSuCo Workshop at SC 2017 in Denver, CO.

  • October 2017: GT Research Horizons article on our DARPA CHIPS Project.

  • September 2017: Dr. Krishna gave a Tutorial on Garnet2.0 in the gem5 workshop at the ARM Research Summit 2017 in Cambridge, UK.
  • September 2017: Paper on Swaps to emulate VCs in NoCs accepted in NoCArc 2017. Congrats Mayank!
  • August 2017: Excited to be a part of the Georgia Tech team that has won the DARPA CHIPS award for enabling low-cost plug-and-play “chiplet”-based hardware design.
  • July 2017: Paper on micro-switch NoC for Neural Network accelerators accepted in NOCS 2017. Congrats Hyoukjun and Anand!
  • July 2017: Dr. Krishna taught a week-long course on On-Chip Networks at the ACACES 2017 Summer School in Fiuggy, Italy.
  • June 2017:  Book Release – On Chip Networks, Second Edition, Morgan Claypool Publishers, co-authored by Natalie Enright Jerger, Tushar Krishna and Li-Shiuan Peh, released at ISCA 2017. Purchase the book from here | Free for universities with academic licenses.
  • June 2017:  Paper on DVFS over SMART NoCs accepted in ICCAD 2017. Congrats Monodeep!
  • May 2017: Dr. Krishna published an essay on The Future of Network-on-Chip (NoC) Architectures in the Circuit Cellar Magazine.
  • May 2017: Congrats to Aniruddh Ramrakhyani and Mayank Parasar for receiving their MS in ECE degrees at the GT Commencement.
  • May 2017: Dr. Krishna gave at talk at Baidu Research’s Silicon Valley AI Lab
  • April 2017: Dr. Krishna gave a talk at TAMU CESG Seminar Series
  • February 2017: SMART-LEES paper on CAD tool for III-V interconnects wins Best Paper Award at DATE 2017!
  • January 2017: OpenSMART paper accepted in ISPASS 2017. Congrats Hyoukjun!
  • November 2016: SMART-LEES paper on CAD tool for III-V interconnects accepted in DATE 2017.
  • October 2016: Dr. Krishna giving a talk at CMU CALCM Seminar Series.
  • October 2016: Static Bubble NoC paper on deadlock-recovery accepted in HPCA 2017. Congrats Aniruddh!
  • October 2016: Garnet2.0 released!
  • Aug 2016: Welcome to ECE undergraduate Fei Wu.
  • April 2016: Welcome Aniruddh Ramrakhyani as the lab’s first MS student.
  • Jan 2016: Welcome Hyoukjun Kwon (CS) and Mayank Parasar (ECE) as the lab’s first two PhD students.
  • Aug 2015: Dr. Tushar Krishna starts at Georgia Tech.

The whole is greater than the sum of its parts