PACT 2026October 19–22, 2026

Program at a Glance

Monday, October 19, 2026

Workshops and tutorials day. Rooms and detailed times are coming soon unless noted below.

Time What
Morning Workshop: 1st Workshop on ML for Assisting Code Quality (MLAC).
Chair: Jay Lofstead, Sandia National Laboratories.
Morning Tutorial: CEDR: A Holistic Software and Hardware Design Environment for Hardware Agnostic Application Development and Deployment on FPGA-Integrated Heterogeneous Systems.
Presenters: Serhan Gener, Umut Suluhan, Ali Akoglu.
Morning Tutorial: SODA Synthesizer: Accelerating Artificial Intelligence Applications with an End-to-End Silicon Compiler.
Presenters: Nicolas Bohm Agostini, Vito Giovanni Castellana, Fabrizio Ferrandi, Serena Curzel, Ankur Limaye, Antonino Tumeo.
All day Workshop: 1st National Science Data Fabric Summit.
Organizers: Michela Taufer, University of Tennessee, Knoxville; Valerio Pascucci, University of Utah.
Afternoon Workshop: 1st LACS — Learning-Augmented Compilers & Systems.
Organizers: Eun Jung (EJ) Park, Riyadh Baghdadi, Joseph Manzano, Keren Zhou.
Afternoon Tutorial: Chameleon: An Experimental Testbed for Parallel, Heterogeneous, and AI-Accelerated Computing.
Presenters: Kate Keahey, Marc Richardson.
Afternoon Tutorial: Reproducible Benchmarking for High-Performance Computing Applications.
Presenters: Olga Pearce, Gregory Becker, Doug Jacobsen, Stephanie Brink
5:00pm-6:00pm Conference reception.

Tuesday, October 20, 2026

Time What
Morning Opening, keynote, and technical sessions: Coming soon.
Midday Lunch/break: Coming soon.
Afternoon Technical sessions: Coming soon.
Evening Poster presentation and ACM Student Research Competition.
Time and room: Coming soon.

Wednesday, October 21, 2026

Time What
Morning Keynote and technical sessions: Coming soon.
Midday Lunch/break: Coming soon
Afternoon Technical sessions: Coming soon.
Evening Conference banquet. Location to be defined.

Thursday, October 22, 2026

Time What
Morning Technical sessions and ACM SRC activities: Coming soon.
Midday Closing session: Coming soon.


Keynotes

Andrew A Chien Tuesday, October 20, 2026

UpDown: A Supercomputer co-designed for Graph Computing and Data Transformation

Andrew A Chien

Univ of Chicago and Argonne National Lab
Chicago UpDown Computing, Inc.

Supercomputers and AI compute are ill-suited for sparse and data-intensive computations because they are optimized for maximum dense matrix "FLOPS". We have designed the UpDown System – optimized for graph computing, streaming data ingestion and transformation as well as high-level programming. The result outperforms conventional CPU/GPU-based systems by 10-100x on an ISO-power basis.

UpDown's radical micro-architecture unleashes fine-grained parallelism: 1-cycle thread creation and management, 1-cycle messages. This enables efficient computation on 10-instruction thread invocations. Further, software-controlled split-transaction DRAM access unlocks the power of HBM's massive memory bandwidth. For irregular applications, UpDown datapath efficiency is 10x greater. UpDown performance on skewed-graph computations exceeds multicore CPU's (>100x) and GPU's (20-60x) in single-node configuration. Updown performance scales to 1,000 and 10,000-fold speedup on BFS, Pagerank, Triangle Count, K-truss and more.

UpDown data ingestion exceeds 5 billion/records/s/node (1000x CPU-based databases), reaching 100 trillion records/s. It enables a new class of streaming analytics and complex workflows. UpDown enables high level programming with a global address space, and a flexible map-reduce framework (KVMSR) coupled with an event-driven language (UDWeave). This enables easy vertex, edge-centric programming, and fits well for other data-parallel models such as relational/graphDB, sparse matrixes, and more.

UpDown was created under funding from IARPA's AGILE program, and is being commercialized by Chicago UpDown Computing, Inc. (www.chupdown.com).

Bio

Andrew A Chien is the William Eckhardt Distinguished Service Professor of Computer Science at the University of Chicago and Senior Scientist at Argonne National Laboratories. Chien led the IARPA funded "UpDown System Project", designing breakthrough scalable graph analytics systems and is now Founder and President of Chicago UpDown Computing, Inc. (www.chupdown.com). He has led the Zero-carbon Cloud project since 2015, and is known for his research on datacenters, renewable energy and sustainability, cloud resource management and software, and large-scale system architecture. Chien has received numerous recognitions for research. Dr. Chien currently serves on the NSF CISE Advisory Committee and DARPA ISAT. He is a Fellow of the ACM, IEEE, and AAAS. He served as EiC of Communications of the ACM, 2017-2022, and Vice President of Research at Intel Corporation from 2005-2010. He served as SAIC Chair Professor of University of California, San Diego (1998-2005) and as faculty at the University of Illinois (1990-98). He received BS, MS, and PhD degrees from the Massachusetts Institute of Technology.

Josep Torrellas Wednesday, October 21, 2026

Toward Accelerator-Centric Computing

Josep Torrellas

Thomas M. Siebel Chair in Computer Science
Director, SRC JUMP 2.0 ACE Center for Evolvable Computing
University of Illinois, Urbana-Champaign
iacoma.cs.uiuc.edu/josep/torrellas.html

Given current energy-consumption trends, there is ample consensus that we will have to move much of the computation to hardware accelerators. This is because accelerators are the most energy-efficient platforms. However, from the evidence of past efforts in this direction, architecting an accelerator-centric computing environment looks very challenging. It is unclear what architectural designs and software advances will really enable this new paradigm. In this talk, I will outline our vision of the hardware and software needed for a successful accelerator-centric computing environment, and some of the efforts that we are doing in this direction.

Bio

Josep Torrellas is the Thomas M. Siebel Chair in Computer Science at the University of Illinois, Urbana-Champaign (UIUC). He is the Director of the ACE Center for Evolvable Computing (an SRC/DARPA JUMP 2.0 Center), past Co-Leader of an Intel Strategic Research Alliance (ISRA) on Computer Security, and past Director of the Illinois-Intel Parallelism Center (I2PC). His research interests are multiprocessor computer architectures and parallel computing. Some of his contributions include thread-level speculation (TLS) architectures, the Bulk Multiprocessor concept, deterministic record and replay mechanisms, process variation mitigation techniques, and hardware defenses against speculative execution attacks. In addition, he has contributed to several experimental multiprocessor designs such as IBM's PERCS Multiprocessor, Intel's Runnemede Extreme-Scale Multiprocessor, Illinois Cedar, and Stanford DASH.

Torrellas has received the IEEE Computer Society (CS) Harry H. Goode Memorial Award, the UIUC Daniel C. Drucker Eminent Faculty Award, the UIUC Campus Award for Excellence in Graduate Student Mentoring, the IEEE CS Edward J. McCluskey Technical Achievement Award, and was a Willett Faculty Scholar at UIUC. He is an IEEE CS Golden Core Member, and a Fellow of IEEE, ACM, and AAAS. He was the Chair of the IEEE Technical Committee on Computer Architecture (TCCA). He has served in the Board of Directors of the Computing Research Association (CRA) and has been a Council Member of CRA's Computing Community Consortium (CCC). He was a member of the U.S. National Academies Board on Army Research and Development. He serves in the International Roadmap for Devices and Systems (IRDS). Torrellas has graduated 53 PhDs. He received a PhD from Stanford University.

Mary Hall Thursday, October 22, 2026

Tiles, Bricks, and Layouts: How Aggregate Data Abstractions Aid in Optimizing Data Movement

Mary Hall

Professor, Kahlert School of Computing, University of Utah

Data movement is the dominant execution and energy cost across the application workloads in data centers and supercomputers. Programming at the tile level has become a popular strategy for optimizing data movement for both deep learning and general structured grids, using Triton, cuTile, bricks, and fine-grained data blocks. Expressing hierarchical data and thread layouts, mostly designed with matrix processors in mind, facilitates automatic code generation that further raises the level of abstraction in such code. In this talk, we will describe prior work on BrickLib supporting fine-grained data blocks and active research on LEGO for hierarchical data and thread layout. We will connect these concepts with emerging hardware features and future demands on programming systems to reduce data movement.

Bio

Mary Hall is a Professor and former Director of the Kahlert School of Computing at University of Utah. Her research focuses on high-performance computing, compiler optimizations and code generation for novel and emerging hardware, and performance tuning. She has served on the Board of Directors of the Computing Research Association since 2015, and she is currently its Vice Chair. She is an ACM Distinguished Scientist and an IEEE Fellow.

Important Dates and Deadlines

Conference Papers:

  • Abstract submission deadline: April 17, 2026 (extended to April 23 2026)
  • Paper submission deadline: April 24, 2026 (extended to April 30, 2026)
  • Rebuttal Period: July 12-16, 2026 (changed to July 19-23, 2026)
  • Author Notification August 5, 2026
  • Artifact submission: August 3-10, 2026 (AoE)
  • Camera ready papers: October 2, 2026

ACM SRC:

  • Abstract Registration Deadline: August 17, 2026
  • Abstract Submission Deadline: August 21, 2026

Conference: October 19–22, 2026


Sponsors

Platinum

Silver

Bronze

Supporters


Previous PACTs

Earlier PACTs