Release Notes » Release 3.4.0 (2022/05/23)

Taskflow 3.4.0 is the 5th release in the 3.x line! This release includes several new changes, such as pipeline parallelism, deadlock-free execution methods, documentation, examples, and unit tests.

Download

Taskflow 3.4.0 can be downloaded from here.

System Requirements

To use Taskflow v3.4.0, you need a compiler that supports C++17:

• GNU C++ Compiler at least v8.4 with -std=c++17
• Clang C++ Compiler at least v6.0 with -std=c++17
• Microsoft Visual Studio at least v19.27 with /std:c++17
• AppleClang Xcode Version at least v12.0 with -std=c++17
• Nvidia CUDA Toolkit and Compiler (nvcc) at least v11.1 with -std=c++17
• Intel C++ Compiler at least v19.0.1 with -std=c++17
• Intel DPC++ Clang Compiler at least v13.0.0 with -std=c++17 and SYCL20

Taskflow works on Linux, Windows, and Mac OS X.

Release Summary

This release enhances our task-parallel pipeline programming model and executor methods, supplied with several new examples and unit tests.

New Features

Taskflow Core

• Improved the pipeline performance using vertical stack optimization
• Added tf::ScalablePipeline to allow programming variable lengths of pipes
• Added tf::Runtime::run_and_wait to allow spawning a subflow
• Added tf::Executor::run_and_wait to allow running taskflows from a worker
• Added an example of attaching data to a task (examples/attach_data.cpp)
• Added an example of text processing pipeline (examples/parallel_text_pipeline.cpp)
• Added an example of graph processing pipeline (examples/parallel_graph_pipeline.cpp)
• Added an example of taskflow processing pipeline (examples/parallel_taskflow_pipeline.cpp)
• Added an example of running a task graph from a worker (examples/run_and_wait.cpp)

cudaFlow

• Added tf::cudaStream as a move-only, RAII-styled wrapper over a native CUDA stream
• Added tf::cudaEvent as a move-only, RAII-styled wrapper over a native CUDA event

syclFlow

There is no update on syclFlow in this release.

Utilities

• Removed serializer to improve compilation speed

Bug Fixes

• Fixed the compilation error due to non-portable include of immintrin.h (#371)
• Fixed the compilation error due to using old version of doctest (#372)
• Fixed the infinite loop bug due to unexpected share states in pipeline (#402)

If you encounter any potential bugs, please submit an issue at issue tracker.

Breaking Changes

• Replaced tf::Runtime::run with tf::Runtime::run_and_wait to comply with tf::Executor::run_and_wait

Deprecated and Removed Items

There are no deprecated items in this release.

Documentation

• Revised Executor

  • Added Execute a Taskflow from an Internal Worker Cooperatively

• Revised Task-parallel Pipeline

• Revised Learning from Examples

  • Added Text Processing Pipeline
  • Added Graph Processing Pipeline
  • Added Taskflow Processing Pipeline

• Added Scalable Task-parallel Pipeline

Miscellaneous Items

We have published Taskflow in the following venues:

• Dian-Lun Lin and Tsung-Wei Huang, "Accelerating Large Sparse Neural Network Inference using GPU Task Graph Parallelism," IEEE Transactions on Parallel and Distributed Systems (TPDS), 2022
• Cheng-Hsiang Chiu and Tsung-Wei Huang, "Composing Pipeline Parallelism using Control Taskflow Graph," ACM International Symposium on High-Performance Parallel and Distributed Computing (HPDC), Minneapolis, Minnesota, 2022
• Cheng-Hsiang Chiu and Tsung-Wei Huang, "Efficient Timing Propagation with Simultaneous Structural and Pipeline Parallelisms," ACM/IEEE Design Automation Conference (DAC), San Francisco, CA, 2022

Please do not hesitate to contact Dr. Tsung-Wei Huang if you intend to collaborate with us on using Taskflow in your scientific computing projects.