An official website of the United States government
Task graphs have been studied for decades as a foundation for scheduling irregular parallel applications and incorporated in many programming models including OpenMP. While many high-performance parallel libraries are based on task graphs, they also have additional scheduling requirements, such as synchronization within inner levels of data parallelism and internal blocking communications. In this paper, we extend task-graph scheduling to support efficient synchronization and communication within tasks. Compared to past work, our scheduler avoids deadlock and oversubscription of worker threads, and refines victim selection to increase the overlap of sibling tasks. To the best of our knowledge, our approach is the first to combine gang-scheduling and work-stealing in a single runtime. Our approach has been evaluated on the SLATE high-performance linear algebra library. Relative to the LLVM OMP runtime, our runtime demonstrates performance improvements of up to 13.82%, 15.2%, and 36.94% for LU, QR, and Cholesky, respectively, evaluated across different configurations related to matrix size, number of nodes, and use of CPUs vs GPUs.
more »
« less
Saturn: An Optimized Data System for Multi-Large-Model Deep Learning Workloads
Large models such as GPT-3 and ChatGPT have transformed deep learning (DL), powering applications that have captured the public's imagination. Such models must be trained on multiple GPUs due to their size and computational load, driving the development of a bevy of model parallelism techniques and tools. Navigating suchparallelismchoices, however, is a new burden for DL users such as data scientists, domain scientists, etc., who may lack the necessary systems knowhow. The need formodel selection, which leads to many models to train due to hyper-parameter tuning or layer-wise finetuning, compounds the situation with two more burdens:resource apportioningandscheduling.In this work, we unify these three burdens by formalizing them as a joint problem that we call SPASE: Select a Parallelism, Allocate resources, and Schedule. We propose a new information system architecture to tackle the SPASE problem holistically, exploiting the performance opportunities presented by joint optimization. We devise an extensible template for existing parallelism schemes and combine it with an automated empirical profiler for runtime estimation. We then formulate SPASE as an MILP. We find that direct use of an MILP-solver is significantly more effective than several baseline heuristics. We optimize the system runtime further with an introspective scheduling approach. We implement all these techniques into a new data system we call Saturn. Experiments with benchmark DL workloads show that Saturn achieves 39-49% lower model selection runtimes than current DL practice.
more »
« less
- Award ID(s):
- 1942724
- PAR ID:
- 10512372
- Publisher / Repository:
- VLDB Endowment
- Date Published:
- Journal Name:
- Proceedings of the VLDB Endowment
- Volume:
- 17
- Issue:
- 4
- ISSN:
- 2150-8097
- Page Range / eLocation ID:
- 712 to 725
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Task graphs have been studied for decades as a foundation for scheduling irregular parallel applications and incorporated in many programming models including OpenMP. While many high-performance parallel libraries are based on task graphs, they also have additional scheduling requirements, such as synchronization within inner levels of data parallelism and internal blocking communications. In this paper, we extend task-graph scheduling to support efficient synchronization and communication within tasks. Compared to past work, our scheduler avoids deadlock and oversubscription of worker threads, and refines victim selection to increase the overlap of sibling tasks. To the best of our knowledge, our approach is the first to combine gang-scheduling and work-stealing in a single runtime. Our approach has been evaluated on the SLATE high-performance linear algebra library. Relative to the LLVM OMP runtime, our runtime demonstrates performance improvements of up to 13.82%, 15.2%, and 36.94% for LU, QR, and Cholesky, respectively, evaluated across different configurations related to matrix size, number of nodes, and use of CPUs vs GPUsmore » « less
-
Many applications that use large-scale machine learning (ML) increasingly prefer different models for subgroups (e.g., countries) to improve accuracy, fairness, or other desiderata. We call this emerging popular practice learning over groups , analogizing to GROUP BY in SQL, albeit for ML training instead of SQL aggregates. From the systems standpoint, this practice compounds the already data-intensive workload of ML model selection (e.g., hyperparameter tuning). Often, thousands of models may need to be trained, necessitating high-throughput parallel execution. Alas, most ML systems today focus on training one model at a time or at best, parallelizing hyperparameter tuning. This status quo leads to resource wastage, low throughput, and high runtimes. In this work, we take the first step towards enabling and optimizing learning over groups from the data systems standpoint for three popular classes of ML: linear models, neural networks, and gradient-boosted decision trees. Analytically and empirically, we compare standard approaches to execute this workload today: task-parallelism and data-parallelism. We find neither is universally dominant. We put forth a novel hybrid approach we call grouped learning that avoids redundancy in communications and I/O using a novel form of parallel gradient descent we call Gradient Accumulation Parallelism (GAP). We prototype our ideas into a system we call Kingpin built on top of existing ML tools and the flexible massively-parallel runtime Ray. An extensive empirical evaluation on large ML benchmark datasets shows that Kingpin matches or is 4x to 14x faster than state-of-the-art ML systems, including Ray's native execution and PyTorch DDP.more » « less
-
Deep learning (DL) is growing in popularity for many data analytics applications, including among enterprises. Large business-critical datasets in such settings typically reside in RDBMSs or other data systems. The DB community has long aimed to bring machine learning (ML) to DBMS-resident data. Given past lessons from in-DBMS ML and recent advances in scalable DL systems, DBMS and cloud vendors are increasingly interested in adding more DL support for DB-resident data. Recently, a new parallel DL model selection execution approach called Model Hopper Parallelism (MOP) was proposed. In this paper, we characterize the particular suitability of MOP for DL on data systems, but to bring MOP-based DL to DB-resident data, we show that there is no single "best" approach, and an interesting tradeoff space of approaches exists. We explain four canonical approaches and build prototypes upon Greenplum Database, compare them analytically on multiple criteria (e.g., runtime efficiency and ease of governance) and compare them empirically with large-scale DL workloads. Our experiments and analyses show that it is non-trivial to meet all practical desiderata well and there is a Pareto frontier; for instance, some approaches are 3x-6x faster but fare worse on governance and portability. Our results and insights can help DBMS and cloud vendors design better DL support for DB users. All of our source code, data, and other artifacts are available at https://github.com/makemebitter/cerebro-ds.more » « less
-
null (Ed.)Deep learning (DL) is a popular technique for building models from large quantities of data such as pictures, videos, messages generated from edges devices at rapid pace all over the world. It is often infeasible to migrate large quantities of data from the edges to centralized data center(s) over WANs for training due to privacy, cost, and performance reasons. At the same time, training large DL models on edge devices is infeasible due to their limited resources. An attractive alternative for DL training distributed data is to use micro-clouds---small-scale clouds deployed near edge devices in multiple locations. However, micro-clouds present the challenges of both computation and network resource heterogeneity as well as dynamism. In this paper, we introduce DLion, a new and generic decentralized distributed DL system designed to address the key challenges in micro-cloud environments, in order to reduce overall training time and improve model accuracy. We present three key techniques in DLion: (1) Weighted dynamic batching to maximize data parallelism for dealing with heterogeneous and dynamic compute capacity, (2) Per-link prioritized gradient exchange to reduce communication overhead for model updates based on available network capacity, and (3) Direct knowledge transfer to improve model accuracy by merging the best performing model parameters. We build a prototype of DLion on top of TensorFlow and show that DLion achieves up to 4.2X speedup in an Amazon GPU cluster, and up to 2X speed up and 26% higher model accuracy in a CPU cluster over four state-of-the-art distributed DL systems.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.14778/3636218.3636227
