Every Bit Counts

Posit Computing
1. Auflage 2024
ISBN: 978-1-032-73806-2
Verlag: CRC Press

Buch, Englisch, 464 Seiten, Format (B × H): 178 mm x 254 mm, Gewicht: 1016 g

Reihe: Chapman & Hall/CRC Computational Science

Written by one of the foremost experts in high-performance computing and the inventor of Gustafson’s law, Every Bit Counts: Posit Computing explains the foundations of a new way for computers to calculate that saves time, storage, energy, and power by packing more information into every bit than do legacy approaches. Both the AI and HPC communities are increasingly using the posit approach that Gustafson introduced in 2017, which may be the future of technical computing.

What may seem like a dry subject is made engaging by including the human and historical side of the struggle to represent numbers on machines. The book is richly illustrated in full color throughout, with every effort made to make the material as clear and accessible as possible, and even humorous.

Starting with the simplest form of the idea, the chapters gradually add concepts according to stated mathematical and engineering design principles, building a robust tool kit for creating application-specific number systems. There is also a thorough explanation of the Posit™ Standard (2022), with motivations and examples that expand on that terse 12-page document.

Zielgruppe

General and Professional Practice & Development

Autoren/Hrsg.

Gustafson, John L

Fachgebiete

• Mathematik | Informatik EDV | Informatik Informatik
• Mathematik | Informatik Mathematik Algebra Zahlentheorie
• Mathematik | Informatik Mathematik Numerik und Wissenschaftliches Rechnen Angewandte Mathematik, Mathematische Modelle
• Mathematik | Informatik Mathematik Mathematik Allgemein Zahlensysteme
• Mathematik | Informatik Mathematik Mathematik Allgemein Diskrete Mathematik, Kombinatorik
• Mathematik | Informatik Mathematik Numerik und Wissenschaftliches Rechnen Numerische Mathematik
• Mathematik | Informatik EDV | Informatik Programmierung | Softwareentwicklung Algorithmen & Datenstrukturen

Inhaltsverzeichnis

Preface  Acknowledgements  How to Read this Book  1. The Posit Number Format  2. Inaccuracy Done Right  3. More Power!  4. How We Got Into This Mess  5. The Posit Standard, Annotated  6. Preaching About the Quire  7. A Bridge to Sane Floats  8. Logarithmic Systems, The Musical  9. The Fastest Arithmetic Ever?  10. The Minefield Method  11. Matrix Multiply, Visualized  12. Linear Solvers that Suck Less  13. The Future of Posit Computing  Glossary  Answers to Exercises  Appendix A: Posit Code Listings  Appendix B: Float Code Listings  Appendix C: The hypot Function  Appendix D: L Table for 8-Bit Posits  Appendix E: Linear Solver Tests  Appendix F: Linear Solvers  Appendix G: Takum Format  Index

Über Autor(innen)

Prof. John L. Gustafson is a pioneer in high-performance computing, known for a breakthrough in parallel processing (Gustafson’s law) for which he received the inaugural Gordon Bell Prize. He is a former Director at Intel Labs and a former Senior Fellow and Chief Product Architect at AMD. He now leads efforts to replace legacy computer arithmetic with high-efficiency next-generation approaches.