Automatic and transparent offloading of computation to near-data processing units

Code used in my PhD project at the University of Edinburgh.

Folders

• memspeed - Measuring memory access latency and speed across tested devices
• ndoffloadsim - Scheduling simulation of the impact of offloading subtasks to storage (and other kinds of) nodes
• pin-NearMAP - a PIN tool

Third-party symlinks setup

To use this repo, you should set up symlinks at:

• third-party/pin3.18 pointing at a Intel PIN 3.18 linux install root (with subdirs doc, extras, source, ...)
• third-party/spec2017 pointing at a SPEC 2017 linux install root (with subdirs benchspec, bin, config, tools, ...)