Events

XiTAO Hackathon

The online XiTAO Hackathon, organised by LEGaTO researchers Miquel Pericas and Mustafa Abduljabbar from the Computer Systems Unit of Chalmers University of Technology, was held on 28 April 2020.

XiTAO Hackathon

XiTAO is one of the components of the LEGaTO toolchain designed to provide energy efficient execution via intelligent resource management in heterogeneous environments. XiTAO is an experimental runtime based on a small code base that is currently undergoing heavy development. The goal of the hackathon was to put the XiTAO user interface under test by adding more benchmarks to the repository. Each participant selected one application from the public domain and attempted to port it over a four-hour period of time. Most of the participants did not have any prior exposure to the XiTAO runtime library, and a few of them had not used any parallel programming framework before encountering XiTAO. Hence, the participants covered various parts of the developer spectrum. 

Applications that were selected for the porting included N-Queens, floorplan, SparseLU, DenseLU (Rodinia), LULESH, Heartwall (Rodinia), Hotspot (Rodinia), Knapsack and GEMM. Despite the short duration of the hackathon, which included also a 1-hour introduction by Mustafa Abduljabbar, considerable progress was achieved on several of these codes. We expect to add several of these implementations to the XiTAO repository over the coming weeks. The hackathon also allowed to identify and fix several bugs in the runtime implementation. 

The organizers created an online wiki for the participants to add questions and provide feedback on the experience. Among positive points that were listed participants mentioned  “learning about XiTAO and how to use it”, and “participating in a stimulating group activity”. On the could-be-improved side, participants asked to be given more learning materials ahead of the hackathon. The organizers plan to have a follow-up hackathon in the coming weeks to complete the implementation of several of these benchmarks.  

Relevant links: 

XiTAO repository: https://github.com/mpericas/xitao

XiTAO-OmpSs integration: https://github.com/legato-project/nanos-xitao-integration

LEGaTO components: https://legato-project.eu/software/components

LEGaTO integrations: https://legato-project.eu/software/integration