COMP Superscalar

Big Data Distributed Computing Programming Models

COMP Superscalar (COMPSs) is a framework which aims to ease the development and execution of parallel applications for distributed infrastructures, such as Clusters, Clouds and containerized platforms.

Software Author: 

Workflows and Distributed Computing Group

Contact:

Jorge Ejarque (jorge [dot] ejarque [at] bsc [dot] es)

Rosa M. Badia (rosa [dot] m [dot] badia [at] bsc [dot] es)

Support mailing list (support-compss [at] bsc [dot] es)

Software Cost: 

COMP Superscalar is distributed under Apache License version 2

Primary tabs

COMPSs Download Form
Please, fill the following form in order to access this download:
2 + 7 =
Solve this simple math problem and enter the result. E.g. for 1+3, enter 4.
COMPSs VM Download Form
Please, fill the following form in order to access this download:
11 + 8 =
Solve this simple math problem and enter the result. E.g. for 1+3, enter 4.

2.9 (Latest Version)

COMP Superscalar version 2.9 (Jasmine) Release date: June-2021

Release Notes

New features

  • Runtime:
    • Support for nested tasks in agents deployment
    • New application time out functionality to enable a controlled finalization of applications before the wall_clock_limit.
    • New flags to simplify application debugging (--keep_workingdir, --gen_coredump).
    • Support for loading the application environment from scripts.
    • Support for tracing in agents environment.
    • Partial support for OSX systems.
  • Python:
    • Support for Optional parameters and default values.
    •  Support for monitoring task status from Jupyter notebooks.
    • Memory profile enabled
    •  Support for Python workers cache.
    •  Support for dynamic number of returns override at task invocation.
  • DDS:
    • New methods and optimizations in DDS class.

Improvements:

  • Enabling the pass of extra flags for the queue system flags from enqueue_compss.
  • Support for multiple data layout in MPI tasks.
  • Improvements in tracing system. More events and cfgs.
  • Enabling a flag to change extrae configuration file for python processes.
  • Configuration files for Barbora system.
  • Several Bug fixes.

Known Limitations

  • OSX support is limited to Java and Python2 without CPU affinity (require to execute with --cpu_affinity=disable)
  • Reduce operations can consume more disk space than the manually programmed n-ary reduction
  • Objects used as task parameters must be serializable.
  • Tasks that invoke Numpy and MKL may experience issues if a different MKL threads count is used in different tasks. This is due to the fact that MKL reuses  threads in the different calls and it does not change the number of threads from one call to another. This can be also happen with other libraries implemented with OpenMP.
  • C++ Objects declared as arguments in a coarse-grain tasks must be passed in the task methods as object pointers in order to have a proper dependency management.
  • Master as worker is not working for executions with persistent worker in C++.
  • Coherence and concurrent writing in parameters annotated with the "Concurrent" direction must be managed by the underlaying distributed storage system.
  • Delete file calls for files used as input can produce a significant synchronization of the main code.
  • Defining a parameter as OUT is only allowed for files and collection files.
  • There is an issue with hwloc and Docker which could affect to python mpi workers. Fixing it require to upgrade the hwloc version used by the MPI runtime.

For further information please refer to COMPSs Documentation

COMPSs for Supercomputers
Check Installation manual for details about how to install from the repository

Read this document before downloading the VM image: COMPSs VM Instructions

Docker Image pull command:

docker pull compss/compss:2.9

COMPSs Docker Image

Old Versions

2.8

2.7

2.6

2.5

2.4

2.3

2.2

2.1

2.0