Hello, and welcome to the summer 2026 edition of the Chapel newsletter! This quarter's report features news about a number of new Chapel publications, along with the usual round-up of talks, blog posts, and other community events. So without further ado, let's dive in, starting with our summer release:
Chapel 2.9
Version 2.9 of Chapel was released on June 18, 2026 and it continued our recent focus of resolving user issues (26!) and improving tools for Chapel development such as the language server, Mason package manager and its ecosystem, chplcheck linter, chpldoc, chapel-py, and c2chapel. This release introduced the ability to dynamically load Chapel libraries that use parallel and distributed features for the first time, and it also made significant improvements to Chapel's long-neglected union types. In addition, Chapel 2.9 added support for LLVM 22, LLDB 22, and CUDA 13.
To learn about Chapel 2.9 in more detail, be sure to read its release announcement on the Chapel blog!
New Chapel Publications
Professor Wu Feng's SyNeRGy Lab at Virginia Tech has been busy lately, with three new Chapel-related papers recently accepted for publication:
-
To ChAI or Not To ChAI: Chapel+AI for Training and Inference by Mohamed Abdelnaby, Paul Sathre, and Wu Feng, to be presented at the 30th IEEE High Performance Extreme Computing Conference (HPEC 2026) on Thursday, September 17, 2026.
-
Chapel for Parallel AND Distributed GPU Computing: A Case Study with Jaccard Similarity by Paul Sathre and Wu Feng, to be presented at ICPP '26 — The 55th International Conference on Parallel Processing during the week of September 28th (details TBA).
-
Beyond the HBM Wall: Coherent-Memory Oversubscription for AI Workloads on NVIDIA Grace Hopper by Mohamed Abdelnaby and Wu Feng, to be presented at the State of Practice in Deploying Supercomputers with NVIDIA Superchips Workshop (SPIN-NVSC) at ICPP '26 on September 28, 2026. This workshop is being organized by the NVIDIA Supercomputing Users Group (NVSUG).
Meanwhile, Shreyas Khandekar described Chapel and its application to Arkouda and ChOp in A Survey of Distributed Asynchronous Many-Task Models and Their Applications by Joseph Schuchart, Patrick Diehl, et al., to be published in an upcoming issue of ACM Computing Surveys (CSUR). At present, a preview is available, but please note that its tables have some unfortunate typographical issues (like missing checkmarks) that will be addressed in the final publication.
Finally, Ian Henriksen from the Qthreads team at Sandia has been leading a collaboration studying Chapel on RISC-V processors, which recently resulted in the paper Porting and Benchmarking Chapel on Emerging RISC-V Hardware: an HPC Viability Study, currently under review. In the meantime, a preprint is available on arXiv.
1 Trillion Row Challenge (1TRC)
Chapel intern Eric Vo (UC Berkeley) kicked off his summer by tackling the trillion row challenge in Arkouda and Chapel. Read about his experiences reading and processing 2.4 terabytes of Parquet data in less than a minute in his article on the Chapel blog.
Chapel Deep-Dive Talks
After knocking out the trillion row challenge, Eric gave a deep-dive talk on his summer project, enhancing the Parquet Mason package so that it could support Arkouda, replacing the bespoke Parquet support it has traditionally used. You can hear about Eric's project and experiences in his deep-dive talk, Migrating Arkouda's Parquet Module to Mason.
Earlier in the summer, Daniel Fedorin presented a deep-dive entitled chapel-py: Python bindings for Chapel's compiler, which provided an overview of the core library that supports many tools, such as Chapel's language server and VSCode support.
To propose future deep-dive sessions, add a comment to this GitHub discussion, and be sure to keep track of upcoming sessions by subscribing to the Chapel community calendar.
Talks
On August 17th, Michelle Strout gave a keynote presentation at the ORNL Software and Data Expo (OSDX 2026) entitled Solving problems with software in a post-AI world. Among other topics, she summarized a recent collaboration with researchers at Oak Ridge National Laboratory in which a Chapel-based tool named FastOTF2 has greatly accelerated their ability to process large data files. This project has been spearheaded by Shreyas Khandekar and Brandon Neth at HPE in collaboration with Adam McDaniel, Oscar Hernandez, Wael Elwasif, and Terry Jones at ORNL. The team has written a paper on this work that's currently under review, but in the meantime, here are a few slides from Michelle's keynote:
Meanwhile, earlier this week, Brad Chamberlain gave an introductory Chapel talk for the Scientific Software Engineering Center in the eScience institute at the University of Washington entitled Chapel: Productive Parallel Programming at all Scales
2000 GitHub stars
Finally, in June, Chapel received its 2000th star on GitHub, a nice milestone to kick off the summer. If you're a Chapel fan and haven't awarded us a star yet, consider heading over to GitHub to help us on our way to 2500.
Upcoming Events
-
On Friday, September 4, Brad will be giving a seminar at Virginia Tech entitled Reflections on 25 Years of Parallel Computing: Hardware transformations, programming incrementalism, and the tenacity of Chapel. If you're in the Blacksburg, VA area, be sure to catch the talk, which will be followed up by an interactive tutorial for Va Tech students.
-
Rosa Badia (BSC) and Brad Chamberlain had a Birds of a Feather session accepted to SC26 entitled The Impact of GenAI on HPC Programming and Programming Models. At this point, it has not yet been scheduled, but will take place sometime November 17–19 at SC26 in Chicago.
-
Finally, Brad recorded a second interview with the My Open Source Experience podcast covering topics like HPSF, AI, and licensing. Stay tuned for more information about when that footage will air.
Resources
That's it for this summer's newsletter! For reference, here's a reminder of some general Chapel resources:
Recurring Public Meetings
- Chapel Project Meetings are every Tuesday 10AM PT
- Chapel Deep-Dives and Demos is a flexible meeting slot on Thursday 10 AM PT
- This slot is available for in-depth live discussion on specific topics, demos of ongoing work, or anything you may want to talk about with the community. Check upcoming discussion topics and dates, or propose your topic here!