Parallel Models: Difference between revisions

From RCSWiki
Jump to navigation Jump to search
No edit summary
 
Line 10: Line 10:
Working on modern computing clusters (like ARC or Compute Canada sites) involves multiple kinds of resources and scales of parallelism. Disentangling them will require some discussion of parallel models and resource mappings.
Working on modern computing clusters (like ARC or Compute Canada sites) involves multiple kinds of resources and scales of parallelism. Disentangling them will require some discussion of parallel models and resource mappings.


<ui>
<ol>
<li>Serial Computation<\li>
<li>Serial Computation</li>
<li>Shared Memory Parallelism<\li>
<li>Shared Memory Parallelism</li>
<li>Distributed Memory Parallelism<\li>
<li>Distributed Memory Parallelism</li>
<li>Job Level Parallelism<\li>
<li>Job Level Parallelism</li>
<li>GPU Parallelism<\li>
<li>GPU Parallelism</li>
<li>Hybrid Parallelism<\li>
<li>Hybrid Parallelism</li>
<\ui>
</ol>


==References==
==References==

Latest revision as of 20:34, 17 March 2021

Parallel work has a long history in practical human endeavours. Rigorous reasoning about it dates back to roughly the 1950s with the development of job-shop and logistical scheduling theories and their implementation on and for parallel computers. A basic working knowledge of parallelism is valuable for learning how to work effectively on a computing cluster.

Introduction

It is easy to identify examples of parallelism in everyday life and scientific work:

  • Coordinated construction of large structures (e.g. Hadrian's Wall)
  • 12 Bakers with 12 Ovens baking 24 loaves of bread in the time it would take one baker to bake 2
  • Multi-channel pipetting
  • Parallel computing

Working on modern computing clusters (like ARC or Compute Canada sites) involves multiple kinds of resources and scales of parallelism. Disentangling them will require some discussion of parallel models and resource mappings.

  1. Serial Computation
  2. Shared Memory Parallelism
  3. Distributed Memory Parallelism
  4. Job Level Parallelism
  5. GPU Parallelism
  6. Hybrid Parallelism

References

Introductory Talk on Parallel Models and Scheduling

Parallelism and Scheduling Slides