eFlows4HPC project aims at providing workflow software stack and an additional set of services to enable the integration of HPC simulations and modelling with big data analytics and machine learning in scientific and industrial applications. The project is also developing the HPC Workflows as a Service (HPCWaaS) methodology that aims at providing tools to simplify the development, deployment, execution and reuse of workflows. The project demonstrates its advances through three application Pillars ...

Provenance registration is becoming more and more important, as we increase the size and number of experiments performed using computers. In particular, when provenance is recorded in HPC environments, it must be efficient and scalable. In this paper, we propose a provenance registration method for scientific workflows, efficient enough to run in supercomputers (thus, it could run in other ...

Session during the Innovative HPC workflows for industry (https://eflows4hpc.eu/event/innovative-hpc-workflows-for-industry/) that describes how Workflow Provenance is recorded with COMPSs: the background on the tools used, how the recording has been designed, and how to use it and inspect metadata.

Name: Matmul GPU Case 1 Cache-ON Contact Person: cristian.tatu@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: Minotauro-MN4

Matmul running on the GPU leveraging COMPSs GPU Cache for deserialization speedup. Launched using 32 GPUs (16 nodes). Performs C = A @ B Where A: shape (320, 56_900_000) block_size (10, 11_380_000)             B: shape (56_900_000, 10)   block_size (11_380_000, 10)             C: shape (320, 10)                block_size ...

Name: Matmul GPU Case 1 Cache-OFF Contact Person: cristian.tatu@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs 3.3 Machine: Minotauro-MN4

Matmul running on the GPU without Cache. Launched using 32 GPUs (16 nodes). Performs C = A @ B Where A: shape (320, 56_900_000) block_size (10, 11_380_000)             B: shape (56_900_000, 10)   block_size (11_380_000, 10)             C: shape (320, 10)                block_size (10, 10) Total dataset size 291 ...

Name: K-Means GPU Cache OFF Contact Person: cristian.tatu@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: Minotauro-MN4

K-Means running on GPUs. Launched using 32 GPUs (16 nodes). Parameters used: K=40 and 32 blocks of size (1_000_000, 1200). It creates a block for each GPU. Total dataset shape is (32_000_000, 1200). Version dislib-0.9

Average task execution time: 194 seconds

Name: K-Means GPU Cache ON Contact Person: cristian.tatu@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: Minotauro-MN4

K-Means running on the GPU leveraging COMPSs GPU Cache for deserialization speedup. Launched using 32 GPUs (16 nodes). Parameters used: K=40 and 32 blocks of size (1_000_000, 1200). It creates a block for each GPU. Total dataset shape is (32_000_000, 1200). Version dislib-0.9

Average task execution time: 16 seconds

Name: Dislib Distributed Training - Cache ON Contact Person: cristian.tatu@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: Minotauro-MN4

PyTorch distributed training of CNN on GPU and leveraging COMPSs GPU Cache for deserialization speedup. Launched using 32 GPUs (16 nodes). Dataset: Imagenet Version dislib-0.9 Version PyTorch 1.7.1+cu101

Average task execution time: 36 seconds

Name: Dislib Distributed Training - Cache OFF Contact Person: cristian.tatu@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: Minotauro-MN4

PyTorch distributed training of CNN on GPU. Launched using 32 GPUs (16 nodes). Dataset: Imagenet Version dislib-0.9 Version PyTorch 1.7.1+cu101

Average task execution time: 84 seconds

Lysozyme in water full COMPSs application run at MareNostrum IV, using full dataset with two workers

PyCOMPSs implementation of Probabilistic Tsunami Forecast (PTF). PTF explicitly treats data- and forecast-uncertainties, enabling alert level definitions according to any predefined level of conservatism, which is connected to the average balance of missed-vs-false-alarms. Run of the Kos-Bodrum 2017 event test-case with 1000 scenarios, 8h tsunami simulation for each and forecast calculations for partial and full ensembles with focal mechanism and tsunami data updates.

PyCOMPSs implementation of Probabilistic Tsunami Forecast (PTF). PTF explicitly treats data- and forecast-uncertainties, enabling alert level definitions according to any predefined level of conservatism, which is connected to the average balance of missed-vs-false-alarms. Run of the Boumerdes-2003 event test-case with 1000 scenarios, 8h tsunami simulation for each and forecast calculations for partial and full ensembles with focal mechanism and tsunami data updates.

Name: Random Forest Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: MareNostrum4 This is an example of Random Forest algorithm from dislib. To show the usage, the code generates a synthetical input matrix. The results are printed by screen. This application used dislib-0.9.0

Name: Lanczos SVD Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: MareNostrum4

Lanczos SVD for computing singular values needed to reach an epsilon of 1e-3 on a matrix of (150000, 150). The input matrix is generated synthetically. This application used dislib-0.9.0

Name: Word Count Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

Wordcount is an application that counts the number of words for a given set of files.

To allow parallelism the file is divided in blocks that are treated separately and merged afterwards.

Results are printed to a Pickle binary file, so they can be checked using: python -mpickle result.txt

This example also shows how to manually add input or ...

Name: TruncatedSVD (Randomized SVD) Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs Machine: MareNostrum4

TruncatedSVD (Randomized SVD) for computing just 456 singular values out of a (3.6M x 1200) size matrix. The input matrix represents a CFD transient simulation of aire moving past a cylinder. This application used dislib-0.9.0

Name: Java Wordcount Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

Wordcount application. There are two versions of Wordcount, depending on how the input data is given.

Version 1

''Single input file'', where all the text is given in the same file and the chunks are calculated with a BLOCK_SIZE parameter.

Version 2

''Multiple input files'', where the text fragments are already in different files under ...

Name: Increment Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

Increment is an application that takes three different values and increases them a number of given times.

The purpose of this application is to show parallelism between the different increments.

Execution instructions

Usage:

runcompss --lang=python src/increment.py N initValue1 initValue2 initValue3 

where:

• N: Number of times to increase ...

Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

Simple is an application that takes one value and increases it by five units. The purpose of this application is to show how tasks are managed by COMPSs.

Execution instructions

Usage:

runcompss --lang=python src/simple.py initValue 

where:

• initValue: Initial value for counter

Execution Examples

runcompss --lang=python src/simple.py 1 
runcompss
...

Name: K-means Contact Person: support-compss@bsc.es Access Level: Public License Agreement: Apache2 Platform: COMPSs

Description

K-means clustering is a method of cluster analysis that aims to partition ''n'' points into ''k'' clusters in which each point belongs to the cluster with the nearest mean. It follows an iterative refinement strategy to find the centers of natural clusters in the data.

When executed with COMPSs, K-means first generates the input points by means of ...

Name: Word Count Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

Wordcount is an application that counts the number of words for a given set of files.

To allow parallelism every file is treated separately and merged afterwards.

Execution instructions

Usage:

runcompss --lang=python src/wordcount.py datasetPath 

where:

• datasetPath: Absolute path of the file to parse (e.g. /home/compss/tutorial_apps/python/wordcount/data/) ...

Name: SparseLU Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

The Sparse LU application computes an LU matrix factorization on a sparse blocked matrix. The matrix size (number of blocks) and the block size are parameters of the application.

As the algorithm progresses, the area of the matrix that is accessed is smaller; concretely, at each iteration, the 0th row and column of the current matrix are discarded. ...

Name: Matrix multiplication with Objects Contact Person: support-compss@bsc.es Access Level: public License Agreement: Apache2 Platform: COMPSs

Description

Matrix multiplication is a binary operation that takes a pair of matrices and produces another matrix.

If A is an n×m matrix and B is an m×p matrix, the result AB of their multiplication is an n×p matrix defined only if the number of columns m in A is equal to the number of rows m in B. When multiplying A and B, the ...