Multi-Queues Can Be State-of-the-Art Priority Schedulers | Artifacts

Artifacts which contain everything required for replicating experiments presented in Multi-Queues Can Be State-of-the-Art Priority Schedulers paper.

• multiqueue-based-schedulers.zip — the source code which contains an implementation of the schedulers, the Galois library and scripts for running experiments
• multiqueue-based-schedulers_fast.tar.gz & multiqueue-based-schedulers_slow.tar.gz — images for running different experimental setups

See readme for more detail.

All the updates will be available here: https://github.com/npostnikova/mq-based-schedulers.

== Schedulers ==

The implementation of the designed priority schedulers can be found in Galois-2.2.1/include/Galois/WorkList/:

• StealingMultiQueue.h is the StealingMultiQueue.
• MQOptimized/ contains MQ Optimized variants.

== Running Experiments ==

Fast experiments (fast tag) contain pre-loaded heatmaps and perform all thread executions for the default and best (according to heatmaps) versions of worklists. Estimated time: 30-40 hours.

Slow experiments (slow tag) perform heatmap and all-thread executions. Estimated time: about 5-6 days.

Image Setup Options

1. Pulling an image from docker hub:

docker pull npostnikova/mq-based-schedulers:$tag

2. Downloading archive here:

sudo docker load -i mq-based-schedulers_${tag}.tar.gz

Running Container:

sudo docker run --cap-add SYS_NICE -it npostnikova/mq-based-schedulers:$tag

Running Experiments:

nano $MQ_ROOT/set_envs.sh  # Update configuration if needed!
$MQ_ROOT/run_experiments.sh