MIT Fast Code Seminar
Algorithms, Compilers, Accelerators, and Whatever It Takes


The MIT Fast Code Seminar is a seminar that covers the latest research topics in the theory and practice of performance engineering. Topics of interest include, but are not limited to, algorithm design and implementation; techniques for improving parallelism and locality; high-performance programming languages and frameworks; compilers for parallel code; tools for analyzing performance; hardware techniques for improving performance; parallel and concurrent data structures; models and algorithms for emerging technologies; high-performance solutions for databases, operating systems, networking, and artificial intelligence; and just plain clever hacks. To receive seminar announcements, please subscribe to this mailing list.

Videos for most talks are available by request to Linda Lynch.

Fall 2023 Schedule

Tuesday 10/17/2023 2pmRohan YadavStanford UniversityDistributed Sparse Computing in Python
Tuesday 12/5/2023 2pmNishil TalatiUniversity of MichiganHigh-Performance GPU Code Generation for Mining Motifs in Temporal Graphs

Previous Seminars

Thursday 5/11/2023Gal SelaTechnion - Israel Institute of TechnologySize Operation for Concurrent Data Structures
Thursday 4/20/2023Jialin LiNational University of SingaporeThe Case for Network Ordering in Distributed Systems Design
Wednesday 10/5/2022Rezaul ChowdhuryStony Brook UniversityFast Stencil Computations using FFT and Gaussian Approximations
Wednesday 9/28/2022Harsha SimhadriMicrosoft ResearchApproximate Nearest Neighbor Search algorithms for web-scale search and recommendation
Wednesday 9/14/2022David TenchRutgers University Dynamic Graph Connectivity: To Infinity And Beyond
Wednesday 6/1/2022 Hans VandierendonckQueen's University Belfast Memory Locality Optimisations for Graph Processing
Wednesday 4/20/2022Jessica ShiMIT Bridging Theory and Practice in Parallel Subgraph Computations
Wednesday 4/6/2022Neil ThompsonMIT How close are algorithms to being optimal?
Wednesday 2/16/2022William KuszmaulMIT Linear Probing Revisited: Tombstones Mark the Demise of Primary Clustering
Wednesday 12/8/2021Yuanhao WeiCMU Multi-point Queries on Concurrent Data Structures
Wednesday 11/17/2021Naama Ben-DavidVMware ResearchAlgorithms for Practical Distributed Agreement
Wednesday 11/3/2021Irene ZhangMicrosoft ResearchThe Demikernel Datapath OS Architecture for Microsecond-scale Datacenter Systems
Wednesday 10/13/2021Jan VitekNortheastern UniversityProductivity and Performance Reconciled: The Julia Story
Wednesday 5/12/2021Daniel LemireUniversity of QuebecParsing numbers at a gigabyte per second
Wednesday 4/28/2021Xuhao ChenMITSoftware and Hardware Systems for Emerging Graph Algorithms
Wednesday 4/7/2021Helen XuMIT Data Structure Design for Skewed Dynamic Graph Processing
Wednesday 3/31/2021Tim DavisTexas A&M SuiteSparse:GraphBLAS: graph algorithms in the language of sparse linear algebra
Wednesday 3/24/2021Martin Farach-ColtonRutgers University The Algorithmics of Address Translation
Wednesday 3/17/2021Maryam Mehri DehnaviUniversity of TorontoInspecting Irregular Computation Patterns to Generate Fast Code
Wednesday 3/10/2021Sam WestrickCMU Disentanglement: Provably Efficient Parallel Functional Programming (Slides)
Wednesday 2/24/2021Edgar SolomonikUIUCScalable Algorithms for Tensor Computations (Slides)
Monday 1/25/2021 Ariful AzadIndiana UniversityComputational Building Blocks for Machine Learning on Graphs (Slides)
Monday 1/11/2021Xuehai QianUSC High Performance Graph Mining Systems (Slides)
Monday 12/7/2020Yuanming HuMIT Taichi: A Language for High-Performance Computation on Spatially Sparse Data Structures
Monday 11/30/2020Scott BeamerUC Santa Cruz Efficiently Exploiting Low-Activity Factors to Accelerate RTL Simulation
Monday 11/9/2020Bradley KuszmaulGoogle Everyone Loves File: File Storage Service (FSS) in Oracle Cloud Infrastructure
Monday 11/2/2020Michael BenderStony Brook University Filters
Monday 10/26/2020Alex PothenPurdue University Approximation: A Paradigm for Designing Parallel Graph Algorithms
Monday 10/19/2020Rezaul ChowdhuryStony Brook University Automatic Derivation of Efficient Parallel Recursive Divide-&-Conquer Algorithms for Dynamic Programs
Monday 9/28/2020Michael MahoneyICSI and UC Berkeley ADAHESSIAN: An Adaptive Second Order Optimizer for Machine Learning
Monday 9/21/2020Franz FranchettiCarnegie Mellon University SPIRAL's Operator Language: From Textbook Math to High Performance - With Correctness Guarantees
Monday 8/10/2020James DemmelUC Berkeley Communication-avoiding algorithms for linear algebra, machine learning and beyond
Monday 8/3/2020Kathy YelickUC Berkeley/
Lawrence Berkeley
National Lab
Genomic Analysis and Learning at Scale: Mapping Irregular Computations to Advanced Architectures
Monday 7/27/2020Larry Rudolph and
Steven Martin
Two Sigma Investments, LP/
QuickQuery: GPU-Based Approximate Query Processing for Sub-Second Exploration at Scale
Monday 7/20/2020Stephen ChouMIT Format Abstractions for Sparse Tensor Algebra Compilation
Monday 7/13/2020Michael Axtmann and
Peter Sanders
Karlsruhe Institute
of Technology
Engineering Scalable Parallel Sorting Algorithms
Monday 6/29/2020Umit CatalyurekGeorgia Tech Fast graph analytics on heterogenous and deep-memory architectures
Monday 6/22/2020Aydin BulucLawrence Berkeley
National Lab/UC Berkeley
Sparse Matrices Beyond Solvers: Graphs, Biology, and Machine Learning (Slides)
Monday 6/15/2020Wen-Mei HwuUIUC Fast GPU Code for Graphs (Slides)
Monday 6/8/2020David BaderNew Jersey Institute
of Technology
Solving Global Grand Challenges with High Performance Data Analytics
Monday 6/1/2020John OwensUC Davis Dynamic Data Structures on the GPU
Monday 5/11/2020Alex AikenStanford Program Optimization for Machine Learning
Monday 5/4/2020Ariya ShajiiMIT Seq: a high-performance language for bioinformatics
Monday 4/20/2020Neil Thompson
& Yash Sherry
MIT How fast are Algorithms Improving?
Tuesday 2/18/2020S. Tucker TaftAdaCore Safe Parallel Programming -- ParaSail, Ada 202X, OpenMP, and Rust
Monday 11/25/2019Bruce MaggsDuke University and
Emerald Innovations
A Speed-of-Light Internet Service Provider
Monday 11/18/2019Charith MendisMIT How to Modernize Compiler Technology
Monday 11/4/2019Yunming ZhangMIT GraphIt: A Domain-Specific Language for Writing High-Performance Graph Applications
Monday 10/21/2019Alex ConwayRutgers University SplinterDB: Closing the Bandwidth Gap on NVMe
Monday 9/30/2019Valentin ChuravyMIT Julia: Making dynamic programs run fast
Monday 9/23/2019Riyadh BaghdadiMIT Tiramisu: A Polyhedral Compiler for Dense and Sparse Deep Learning
Monday 9/16/2019Bill DallyNVIDIA Corporation and
Stanford University
Domain-Specific Accelerators
Tuesday 8/27/2019Laxman DhulipalaCarnegie Mellon University Algorithms and Systems for Processing Massive Static and Evolving Graphs
Tuesday 8/20/2019Tao B. SchardlMIT Tapir: Embedding Recursive Fork-Join Parallelism into LLVM's Intermediate Representation
Tuesday 8/6/2019Jeremy KepnerMIT Lincoln Laboratory
Supercomputing Center
Optimal system settings: How to not lose before you begin
Tuesday 7/30/2019I-Ting Angelina LeeWashington University
in St. Louis
Advances in Determinacy Race Detection for Task-Parallel Code
Tuesday 7/23/2019Maurice HerlihyBrown University Speculative Concurrency for Ethereum Smart Contracts
Tuesday 7/16/2019Song HanMIT AutoML for Efficiently Designing Efficient Neural Network Architectures
Tuesday 7/9/2019Fredrik KjolstadMIT The Sparse Tensor Algebra Compiler
Tuesday 6/11/2019Charles LeisersonMIT The Resurgence of Software Performance Engineering


Julian Shun (lead organizer)
Saman Amarasinghe
Adam Belay
Charles Leiserson
Tao B. Schardl