Taskflow: A Lightweight Parallel and Heterogeneous Task Graph Computing System

Huang, Tsung‐Wei; Lin, Dian-Lun; Lin, Chun-Xun; Lin, Yibo

doi:10.1109/tpds.2021.3104255

Cited by 68 publications

(20 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Inspired by the need of parallel CAD algorithms, Pipeflow introduces a new task-parallel pipeline programming model for users to create a pipeline scheduling framework without data abstraction. We establish Pipeflow atop the open-source parallel task graph programming system, Taskflow [15], because it has been successfully adopted by many important CAD projects under the DARPA IDEA program [3,22,36]. In this section, we will first give a brief introduction about Taskflow and then dive into the technical details of Pipeflow.…”

Section: Pipeflowmentioning

confidence: 99%

“…Taskflow is a general-purpose parallel and heterogeneous programming system using modern C++ [15]. Taskflow introduces a new control taskflow graph (CTFG) model that enables end-to-end implementation of task graph parallelism coupled with in-graph control flow.…”

Section: State Of the Art: Taskflowmentioning

confidence: 99%

“…Pipeflow creates a taskflow for each pipeline module task using one condition task and multiple runtime tasks one per line. A runtime task is a task type in Taskflow for users to interact with the executor [15], such as scheduling a task in the graph. The condition task decides which runtime task to run when the pipeline starts.…”

Section: Scheduling Algorithmmentioning

confidence: 99%

“…Consequently, we introduce in this paper Pipeflow, a new task-parallel pipeline programming framework to overcome the limitations of existing ones. We establish Pipeflow atop the open-source task graph programming system, Taskflow, developed by Huang et al [15], to leverage its powerful tasking infrastructure. As Taskflow is being used by several important research projects [3,22,36], building Pipeflow on top will not only benefit existing users but also allow us to gain timely feedback from the Taskflow community.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Pipeflow: An Efficient Task-Parallel Pipeline Programming Framework using Modern C++

Chiu¹,

Huang²,

Guo³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Pipeline is a fundamental parallel programming pattern. Mainstream pipeline programming frameworks count on data abstractions to perform pipeline scheduling. This design is convenient for data-centric pipeline applications but inefficient for algorithms that only exploit task parallelism in pipeline. As a result, we introduce a new task-parallel pipeline programming framework called Pipeflow. Pipeflow does not design yet another data abstraction but focuses on the pipeline scheduling itself, enabling more efficient implementation of task-parallel pipeline algorithms than existing frameworks. We have evaluated Pipeflow on both micro-benchmarks and real-world applications. As an example, Pipeflow outperforms oneTBB 24% and 10% faster in a VLSI placement and a timing analysis workloads that adopt pipeline parallelism to speed up runtimes, respectively.

show abstract

Section: Pipeflowmentioning

confidence: 99%

Section: State Of the Art: Taskflowmentioning

confidence: 99%

Section: Scheduling Algorithmmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Pipeflow: An Efficient Task-Parallel Pipeline Programming Framework using Modern C++

Chiu¹,

Huang²,

Guo³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…In an effort to foster the understanding and development of simulation path heuristics for quantum circuit simulation based on decision diagrams, this section presents an open-source simulation path framework. This framework allows to execute arbitrary simulation paths via the powerful TaskFlow [28] library. Instead of reinventing the wheel and trying to compensate for years of research on tensor contraction methods, the framework includes a push-button flow to employ existing techniques from the tensor network domain while simultaneously providing the means to easily realize dedicated simulation path strategies for decision diagrams.…”

Section: B Related Workmentioning

confidence: 99%

Simulation Paths for Quantum Circuit Simulation with Decision Diagrams

Burgholzer,

Ploier,

Wille

2022

Preprint

View full text Add to dashboard Cite

Simulating quantum circuits on classical computers is a notoriously hard, yet increasingly important task for the development and testing of quantum algorithms. In order to alleviate this inherent complexity, efficient data structures and methods such as tensor networks and decision diagrams have been proposed. However, their efficiency heavily depends on the order in which the individual computations are performed. For tensor networks the order is defined by so-called contraction plans and a plethora of methods has been developed to determine suitable plans. On the other hand, simulation based on decision diagrams is mostly conducted in a straight-forward, i.e., sequential, fashion thus far.In this work, we study the importance of the path that is chosen when simulating quantum circuits using decision diagrams. We propose an open-source framework that not only allows to investigate dedicated simulation paths, but also to re-use existing findings, e.g., obtained from determining contraction plans for tensor networks. Experimental evaluations show that translating strategies from the domain of tensor networks can already yield speedups of up to several orders of magnitude compared to the state of the art. Furthermore, we show, both conceptually and experimentally, that choosing the right simulation path can result in runtimes practically independent of the number of qubitsas opposed to exponential runtimes using the state-of-the-art approach. The proposed framework is publicly available at github.com/cda-tum/ddsim.

show abstract

Programming heterogeneous architectures using hierarchical tasks

Faverge

Furmento

Guermouche

et al. 2023

Concurrency and Computation

View full text Add to dashboard Cite

SummaryTask‐based systems have become popular due to their ability to utilize the computational power of complex heterogeneous systems. A typical programming model used is the Sequential Task Flow (STF) model, which unfortunately only supports static task graphs. This can result in submission overhead and a static task graph that is not well‐suited for execution on heterogeneous systems. A common approach is to find a balance between the granularity needed for accelerator devices and the granularity required by CPU cores to achieve optimal performance. To address these issues, we have extended the STF model in the StarPU runtime system by introducing the concept of hierarchical tasks. This allows for a more dynamic task graph and, when combined with an automatic data manager, it is possible to adjust granularity at runtime to best match the targeted computing resource. That data manager makes it possible to switch between various data layout without programmer input and allows us to enforce the correctness of the DAG as hierarchical tasks alter it during runtime. Additionally, submission overhead is reduced by using large‐grain hierarchical tasks, as the submission process can now be done in parallel. We have shown that the hierarchical task model is correct and have conducted an early evaluation on shared memory heterogeneous systems using the Chameleon dense linear algebra library.

show abstract

Taskflow: A Lightweight Parallel and Heterogeneous Task Graph Computing System

Cited by 68 publications

References 38 publications

Pipeflow: An Efficient Task-Parallel Pipeline Programming Framework using Modern C++

Pipeflow: An Efficient Task-Parallel Pipeline Programming Framework using Modern C++

Simulation Paths for Quantum Circuit Simulation with Decision Diagrams

Programming heterogeneous architectures using hierarchical tasks

Contact Info

Product

Resources

About