Communication Avoiding Algorithms and Programming Patterns

Dear Colleagues,

Today’s machines suffer from a wide gap between memory and processor speed, and they suffer from network latency and bandwidth constraints. This suffering is not expected to go away with new machine generations. For scientific computing, data movements—either on-chip or between nodes—have started to hinder upscaling and dominate our energy bill. They throttle insight through simulations. It is thus a timely endeavour to think about algorithmic reformulations of our tools, and about programming paradigms, rationale and patterns: how can we cast algorithms into implementations that take harsh data movement constraints into account? Such approaches can be called communication avoidance, once we use the label communication for intra- and inter-chip data movements. This comprises the elimination of data volume (through online compression, reduced precision or the re-computation of information), but also covers the reduction of the data exchange frequency, avoiding data movement peaks where all compute entities access precious resources at the same time, hiding movements behind computations, the localisation of data movements, and so forth. This list is not comprehensive. The aim of this Special Issue is to collect novel papers introducing new communication-avoidant ideas, applying “communication avoidance” to applications, improving our understanding of the potential limits of communication avoidance, and predicting its impact on future machine generations. Contributions from the area of numerical simulations are particularly welcome, but authors from other areas are encouraged to contribute, too.

Dr. Tobias Weinzierl
Guest Editor

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Algorithms is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• Communication avoidance
• Scientific computing
• Data compression
• Data-centric programming
• Communications management
• Latency-bound
• Bandwidth-bound