Welcome to the UnoAPI Systems Modules

Note

This is a set of modules we are developing around Intel’s OneAPI initiative. Nothing here is even remotely ready for prime time!

• Preliminaries
  • Motivation
  • Topics
• Parallel Programming Preliminaries
  • Brief History of Data Parallelism
    • Vector machines (Cray)
    • Connection Machine (Thinking Machines)
    • Systolic architectures
    • Loop parallelism in FORTRAN
    • C* and Data-Parallel C efforts
    • CUDA
    • OpenMP
    • oneAPI
  • Parallel Concepts for Humans
    • von Neumann machines and their limits
    • Flynn’s taxonomy
    • Control-memory taxonomy
    • Speedup vs. Efficiency
    • Strong/Weak scaling [new topic]
  • Granularity
    • starvation, deadlock
    • Flooding/Throttling
    • Latency
    • Scheduling [needed for understanding q.submit()]
    • task graphs / dataflow execution / macro-dataflow concept
• Software Engineering
  • Cmake
    • Basic organization of CMake
    • Variables that affect working with DPC++ tooling
    • Separating build from source (viewpathed building)
    • Fetching external dependencies
    • Important external dependencies for C++ 17, 20, and beyond
  • Unit Testing with GoogleTest
  • Continuous Integration
    • GitHub Actions
    • GitLab?
• Modern C++ as a Better C (and C++)
  • Language Features
  • Libraries
    • fmt
    • command-line argument handling
    • logging - spdlog
• Data-Parallel C++
• Selecting and Using Devices
• Performance
• Distributed Computing with OneAPI
• Issues
  • Heading
  • Examples
  • Illustration

Repository

github.com/LoyolaChicagoBooks/unoapi

Releases

github.com/LoyolaChicagoBooks/unoapi/releases/latest

Topics

Indices and tables

• Index

• Module Index

• Search Page