BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/Denver X-LIC-LOCATION:America/Denver BEGIN:DAYLIGHT TZOFFSETFROM:-0700 TZOFFSETTO:-0600 TZNAME:MDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0600 TZOFFSETTO:-0700 TZNAME:MST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260422T000614Z LOCATION:708 DTSTART;TZID=America/Denver:20231112T140000 DTEND;TZID=America/Denver:20231112T173000 UID:submissions.supercomputing.org_SC23_sess438@linklings.com SUMMARY:7th International Workshop on Software Correctness for HPC Applica tions (Correctness '23) DESCRIPTION:Mixed-Precision S/DGEMM Using the TF32 and TF64 Frameworks on Low-Precision AI Tensor Cores\n\nUsing NVIDIA Tensor Cores has enabled the significant acceleration of general matrix multiplication for application s in AI and in high-performance computing. The use of such specialized acc elerators can provide a performance increase between 8x and 20x, albeit wi th a loss in precision. However, high...\n\n\nPedro Valero-Lara (Oak Ridge National Laboratory (ORNL)), Ian Jorquera (Colorado State University), an d Frank Lui and Jeffrey Vetter (Oak Ridge National Laboratory (ORNL))\n--- ------------------\nRethinking Data Race Detection in MPI-RMA Programs\n\n Supercomputers are capable of more and more computations, and nodes formin g them need to communicate even more efficiently with each other. Thus, ne w types of communication models gain traction in the community to promote overlapping communications with computations. For instance, the Message Pa ssin...\n\n\nRadjasouria Vinayagame (Eviden, INRIA); Emmanuelle Saillard ( INRIA); Samuel Thibault (University of Bordeaux); and Van Man Nguyen and M arc Sergent (Eviden)\n---------------------\nAdding Microbenchmarks with S IMD Data Race to DataRaceBench\n\nData race detection tools should find da ta races not only in development builds of applications, but also in optim ized production builds. An architecture-dependent optimization includes v ectorization of the code. At the moment, DataRaceBench does not contain m icrokernels that test for data races i...\n\n\nJoachim Jenke, Kaloyan Igna tov, and Simon Schwitanski (RWTH Aachen University)\n--------------------- \nData Race Detection Using Large Language Models\n\nLarge language models (LLMs) are demonstrating significant promise as an alternate strategy to facilitate analyses and optimizations of high-performance computing progra ms, circumventing the need for resource-intensive manual tool creation. In this paper, we explore a novel LLM-based data race detec...\n\n\nLe Chen (Iowa State University, Lawrence Livermore National Laboratory (LLNL)); Xi anzhong Ding (University of California, Merced); Pei-Hung Lin and Chunhua Liao (Lawrence Livermore National Laboratory (LLNL)); Murali Emani (Argonn e National Laboratory (ANL)); and Tristan Vanderbruggen (Lawrence Livermor e National Laboratory (LLNL))\n---------------------\nRMARaceBench: A Mic robenchmark Suite to Evaluate Race Detection Tools for RMA Programs\n\nPar allel programming models with Remote Memory Access (RMA), such as MPI RMA, OpenSHMEM, and GASPI, allow processes to modify the memory of other proce sses directly. Special care is needed to avoid concurrent conflicting acc esses that lead to data races across processes with undefined behavior. Al t...\n\n\nSimon Schwitanski, Joachim Jenke, Sven Klotz, and Matthias S. Mü ller (Chair for High-Performance Computing, IT Center, RWTH Aachen Univers ity)\n---------------------\nCorrectness Workshop Opening Remarks\n\nIgnac io Laguna (Lawrence Livermore National Laboratory (LLNL)) and Cindy Rubio- González (University of California, Davis)\n---------------------\nHPC Bug s Fest Introduction\n\nEmmanuelle Saillard (French Institute for Research in Computer Science and Automation (INRIA))\n---------------------\nMappin g High-Level Concurrency from OpenMP and MPI to ThreadSanitizer Fibers\n\n High-level parallel programming paradigms like MPI and OpenMP allow expres sing concurrency independent from the execution unit finally executing the code. Most general-purposed data race detection tools perform thread-cen tric analysis with the operating system thread as the execution unit. Thr eadS...\n\n\nJoachim Jenke, Simon Schwitanski, Isabel Thärigen, and Matthi as S. Müller (RWTH Aachen University)\n---------------------\nHighlighting PARCOACH Improvements on MBI\n\nPARCOACH is one of the few verification t ools that relies on a static analysis to detect errors in MPI programs. Fi rst focused on the detection of call ordering errors with collectives, it has recently been extended to detect local concurrency errors in MPI-RMA p rograms. Furthermore, the new version...\n\n\nPhilippe Virouleau and Emman uelle Saillard (French Institute for Research in Computer Science and Auto mation (INRIA)) and Marc Sergent and Pierre Lemarinier (Eviden)\n--------- ------------\nInvestigating the Real-World Applicability of MPI Correctnes s Benchmarks\n\nThe MPI correctness benchmarks MPI-Corrbench and the MPI B ugs Initiative contain standardized test cases of correct and erroneous us e of MPI, allowing MPI correctness tool developers to assess their tools p erformance and guide further development of their checking capabilities. Hence, the correctne...\n\n\nAlexander Hück and Tim Jammer (TU Darmstadt), Joachim Jenke (RWTH Aachen University), and Christian Bischof (TU Darmsta dt)\n---------------------\nToward Correctness Checking of MPI Partitioned Communication in MUST\n\nPartitioned communication introduced with MPI 4. 0 can improve the communication efficiency of hybrid parallel models. It a llows threads on the sender and the receiver side to work on parts of a co mmunication buffer before the communication operation is fully completed. In this presentation, we discu...\n\n\nSimon Schwitanski, Niko Sakic, Joac him Jenke, Felix Tomski, and Marc-André Hermanns (IT Center, RWTH Aachen U niversity)\n---------------------\nImprove and Stabilize Classification Re sults of DataRaceBench\n\nDataRaceBench is a benchmark using small kernel applications to classify the detection capabilities of data race detection tools. During our experiments of applying Archer to the benchmark suite we observed different short-comings. With recently added kernels, the tur n-around time of a basic bench...\n\n\nJoachim Jenke and Simon Schwitanski (RWTH Aachen University)\n---------------------\nCorrectness '23 – Aftern oon Break\n\nTag: Applications, Software Engineering\n\nRegistration Categ ory: Workshop Reg Pass\n\nSession Chairs: Ignacio Laguna (Lawrence Livermo re National Laboratory (LLNL)); Cindy Rubio-González (University of Califo rnia, Davis); and Emmanuelle Saillard (French Institute for Research in Co mputer Science and Automation (INRIA)) END:VEVENT END:VCALENDAR