COMPSs GPU Cache Matrix Multiplication
Version 1

Workflow Type: COMPSs

Name: Matmul GPU Case 1 Cache-ON
Contact Person:
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 (10, 10)
Total dataset size 291 GB.
Version dislib-0.9

Average task execution time: 32 seconds

Click and drag the diagram to pan, double click or use the controls to zoom.

Version History

Version 1 (earliest) Created 22nd Mar 2024 at 12:26 by Cristian Tatu

No revision comments

Frozen Version-1 0fcc18f
help Creators and Submitter
Additional credit

The Workflows and Distributed Computing Team (

Tatu, C. (2024). COMPSs GPU Cache Matrix Multiplication. WorkflowHub.

Views: 321   Downloads: 89

Created: 22nd Mar 2024 at 12:26

Last updated: 25th Mar 2024 at 11:35

Annotated Properties
Topic annotations
help Attributions


Total size: 355 KB
Powered by
Copyright © 2008 - 2024 The University of Manchester and HITS gGmbH