ICFP 2021
Sun 22 - Sat 28 August 2021
Mon 23 Aug 2021 16:15 - 16:30 at ICFP Talks - Session 1
Tue 24 Aug 2021 04:15 - 04:30 at ICFP Talks - Session 1

We present a dataflow model for modelling parallel Unix shell pipelines. To accurately capture the semantics of complex Unix pipelines, the dataflow model is order-aware, i.e., the order in which a node in the dataflow graph consumes inputs from different edges plays a central role in the semantics of the computation and therefore in the resulting parallelization. We use this model to capture the semantics of transformations that exploit data parallelism available in Unix shell computations and prove their correctness. We additionally formalize the translations from the Unix shell to the dataflow model and from the dataflow model back to a parallel shell script. We implement our model and transformations as the compiler and optimization passes of a system parallelizing shell pipelines, and use it to evaluate the speedup achieved on 47 pipelines.

Mon 23 Aug

Displayed time zone: Seoul change

16:00 - 17:30
16:00
15m
Talk
Client-Server Sessions in Linear LogicDistinguished Paper
Research Papers
Zesen Qian Aarhus University, Alex Kavvos University of Bristol, Lars Birkedal Aarhus University
DOI Media Attached
16:15
15m
Talk
An Order-Aware Dataflow Model for Parallel Unix Pipelines
Research Papers
Shivam Handa Massachusetts Institute of Technology, Konstantinos Kallas University of Pennsylvania, Nikos Vasilakis Massachusetts Institute of Technology, Martin C. Rinard Massachusetts Institute of Technology
DOI Media Attached
16:30
15m
Talk
CPS Transformation with Affine Types for Call-By-Value Implicit Polymorphism
Research Papers
Taro Sekiyama National Institute of Informatics, Takeshi Tsukada Chiba University, Japan
DOI Media Attached
16:45
15m
Talk
Getting to the Point: Index Sets and Parallelism-Preserving Autodiff for Pointful Array ProgrammingDistinguished Paper
Research Papers
Adam Paszke Google Research, Daniel D. Johnson Google Research, David Kristjanson Duvenaud University of Toronto, Dimitrios Vytiniotis DeepMind, Alexey Radul Google Research, Matthew J. Johnson Google Research, Jonathan Ragan-Kelley MIT CSAIL, Dougal Maclaurin Google Research
DOI Media Attached
17:00
15m
Talk
Propositions-as-Types and Shared State
Research Papers
Pedro Rocha Nova University of Lisbon, Luís Caires Universidade Nova de Lisboa and NOVA LINCS
DOI Media Attached
17:15
15m
Talk
Efficient Tree-Traversals: Reconciling Parallelism and Dense Data Representations
Research Papers
Chaitanya S. Koparkar Indiana University, Mike Rainey Carnegie Mellon University, Michael Vollmer University of Kent, Milind Kulkarni Purdue University, Ryan R. Newton Facebook
DOI Media Attached

Tue 24 Aug

Displayed time zone: Seoul change

04:00 - 05:30
Session 1Research Papers at ICFP Talks
04:00
15m
Talk
Client-Server Sessions in Linear LogicDistinguished Paper
Research Papers
Zesen Qian Aarhus University, Alex Kavvos University of Bristol, Lars Birkedal Aarhus University
DOI Media Attached
04:15
15m
Talk
An Order-Aware Dataflow Model for Parallel Unix Pipelines
Research Papers
Shivam Handa Massachusetts Institute of Technology, Konstantinos Kallas University of Pennsylvania, Nikos Vasilakis Massachusetts Institute of Technology, Martin C. Rinard Massachusetts Institute of Technology
DOI Media Attached
04:30
15m
Talk
CPS Transformation with Affine Types for Call-By-Value Implicit Polymorphism
Research Papers
Taro Sekiyama National Institute of Informatics, Takeshi Tsukada Chiba University, Japan
DOI Media Attached
04:45
15m
Talk
Getting to the Point: Index Sets and Parallelism-Preserving Autodiff for Pointful Array ProgrammingDistinguished Paper
Research Papers
Adam Paszke Google Research, Daniel D. Johnson Google Research, David Kristjanson Duvenaud University of Toronto, Dimitrios Vytiniotis DeepMind, Alexey Radul Google Research, Matthew J. Johnson Google Research, Jonathan Ragan-Kelley MIT CSAIL, Dougal Maclaurin Google Research
DOI Media Attached
05:00
15m
Talk
Propositions-as-Types and Shared State
Research Papers
Pedro Rocha Nova University of Lisbon, Luís Caires Universidade Nova de Lisboa and NOVA LINCS
DOI Media Attached
05:15
15m
Talk
Efficient Tree-Traversals: Reconciling Parallelism and Dense Data Representations
Research Papers
Chaitanya S. Koparkar Indiana University, Mike Rainey Carnegie Mellon University, Michael Vollmer University of Kent, Milind Kulkarni Purdue University, Ryan R. Newton Facebook
DOI Media Attached