PPoPP 2023
Sat 25 February - Wed 1 March 2023 Montreal, Canada
Tue 28 Feb 2023 10:20 - 10:40 at Montreal 4 - Session 4: Parallelism Chair(s): Gagan Agrawal

The emergence of heterogeneous memory (HM) provides a cost-effective and high-performance solution to memory-consuming HPC applications. Deciding the placement of data objects on HM is critical for high performance. We reveal a performance problem related to data placement on HM. The problem is manifested as load imbalance among tasks in task-parallel HPC applications. The root of the problem comes from being unaware of parallel-task semantics and an incorrect assumption that bringing frequently accessed pages to fast memory always leads to better performance. To address this problem, we introduce a load balance-aware page management system, named \textit{Merchandiser}. Merchandiser introduces task semantics during memory profiling, rather than being application-agnostic. Using the limited task semantics, Merchandiser effectively sets up coordination among tasks on the usage of HM to finish \textit{all} tasks fast instead of only considering any individual task. Merchandiser is highly automated to enable high usability. Evaluating with memory-consuming HPC applications, we show that Merchandiser reduces load imbalance and leads to an average of 17.1% and 15.4% (up to 26.0% and 23.2%) performance improvement, compared with a hardware-based solution and an industry-quality software-based solution.

Tue 28 Feb

Displayed time zone: Eastern Time (US & Canada) change

10:00 - 11:40
Session 4: ParallelismMain Conference at Montreal 4
Chair(s): Gagan Agrawal The Ohio State University
10:00
20m
Talk
OpenCilk: A Modular and Extensible Software Infrastructure for Fast Task-Parallel Code
Main Conference
TB Schardl MIT CSAIL, I-Ting Angelina Lee Washington University in St. Louis, USA
10:20
20m
Talk
Merchandiser: Data Placement on Heterogeneous Memory for Task-Parallel HPC Applications with Load-Balance Awareness
Main Conference
Zhen Xie Argonne National Laboratory, Jie Liu University of California, Merced, Jiajia Li North Carolina State University, Dong Li University of California, Merced
10:40
20m
Talk
Visibility Algorithms for Dynamic Dependence Analysis and Distributed Coherence
Main Conference
Michael Bauer NVIDIA, Elliott Slaughter SLAC National Accelerator Laboratory, Sean Treichler NVIDIA, Wonchan Lee NVIDIA, Michael Garland NVIDIA, Alex Aiken Stanford University
11:00
20m
Talk
Block-STM: Scaling Blockchain Execution by Turning Ordering Curse to a Performance Blessing
Main Conference
Rati Gelashvili Aptos, Alexander Spiegelman Aptos, Zhuolun Xiang Aptos, George Danezis Mysten Labs & University College London, Zekun Li Aptos, Dahlia Malkhi Chainlink Labs, Yu Xia MIT, Runtian Zhou Aptos
11:20
20m
Talk
TL4x - Buffered Durable Transactions on Disk as Fast as in Memory
Main Conference
Gal Assa Technion, Andreia Correia University of Neuchâtel, Pedro Ramalhete Cisco Systems, Valerio Schiavoni University of Neuchatel, Pascal Felber University of Neuchâtel