VIVEKFEST 2024
who it's for
compiler researchers, parallel language researchers, runtime system engineers, high performance computing developers, accelerator compiler engineers, cloud platform engineers, verification and static analysis engineers
topics
middle-end optimizer (AXA/AXAI); SSA and array-SSA timestamps; SMT structural mutation driver; asynchronous actor runtimes; FA-BSP (FABS) user-level tasking; EDDO accelerator affine substitution; portable heterogeneous accelerator model; multigrain parallelizing compiler (OSCAR); OpenSHMEM conveyor aggregation; small-message aggregation optimizations; GPU data-race static analysis (Faial); energy and speedup measurements on Frontier
| title | author | length |
|---|---|---|
| [VIVEKFEST24] Vivek Sarkar (Georgia Tech) | Vivek Sarkar | 35m |
History of compiler and parallel language work with concrete outcomes: a production middle-end optimizer (AXA/AXAI) ships in Fortran compilers since 1996; a… | ||
| Retrieving Unknown SMT Formulas via Structural Mutations | Shuo Ding, Qirun Zhang | 18m |
Claim: unknown SMT formulas often become decidable by simple syntactic edits and running the same solver on those variants in parallel. Method mutates formulas… | ||
| Enabling User-level Asynchronous Tasking in the FA-BSP Model - Case Study:(…) | Akihiro Hayashi, Shubhendra Singhal, Youssef Elmougy, Jiawei Yang | 21m |
Extends FA-BSP into a fine-grained FA-BSP (FABS) runtime that adds actor-based asynchronous messaging and tests enabling user-level async+finish tasking on… | ||
| On the Cloud We Can’t Wait: Asynchronous Actors Perform Even Better on the Cloud | Aniruddha Mysore, Youssef Elmougy, Akihiro Hayashi | 10m |
Comparison of OpenSHMEM (synchronous PGAS) and HClib-actor (asynchronous) on Bale graph microkernels shows asynchronous code runs far faster on cloud VMs with… | ||
| Navigating the Frontier of EDDO-Based Accelerators: Programming and Code(…) | Prasanth Chatarasi | 21m |
Argues that EDDO-style accelerators like IBM AIU trade hardware complexity for much harder software and compiler problems, then shows a concrete codegen fix: a… | ||
| Preliminary Study on Message Aggregation Optimizations for Energy Savings in PGAS(…) | Oscar Hernandez, Aaron Welch, Wendy Poole, Stephen Poole | 16m |
Compares OpenSHMEM conveyor aggregation vs atomic/get/put on Frontier, measuring CPU and memory energy with HPM and visualizing with Scorp and Vampier using… | ||
| A Formal Model for Portable, Heterogeneous Accelerator Programming | Zachary Sullivan, Samuel D. Pollard | 22m |
A formal core language called H-IMP models execution spaces (host, OpenMP, CUDA) and memory spaces (host, CUDA, UVM) with an accessibility relation that… | ||
| Evaluation of Speedup & Energy with Multigrain Parallelizing Compiler | John Pickar, Tohma Kawasumi, Hiroki Mikami, Keiji Kimura, Hironori Kasahara | 17m |
Automatic source-to-source OSCAR compiler applies multigrain parallelism, task-graph scheduling and L3-aware data localization to C/C++/Fortran and compiles… | ||
| Hidden assumptions in static verification of data-race free GPU programs | Tiago Cogumbreiro, Julien Lange | 19m |
Faial is a static analyzer for GPU kernels that checks data-race freedom and delegates checks to an SMT solver, running most kernels in under a minute. On a… | ||
| Scalable Small Message Aggregation on Modern Interconnects | Aaron Welch, Oscar Hernandez, Stephen Poole, Wendy Poole | 24m |
Proposes aggregation contexts as an OpenSHMEM extension that adds one context argument and a quiet flush to defer small remote operations, encoding metadata… | ||
| nothing matches | ||