Further Reading: Why Parallel?

This chapter is about ideas — the hardware shift, the two laws, the taxonomy — more than syntax, so the best sources are a mix of two original papers worth reading in full and a few modern references that put the ideas to work in Fortran. Sources are tagged Tier 1 (works we are confident exist and recommend) and Tier 2 (real and worth seeking; confirm the current edition or URL yourself). Read enough here to make Amdahl's ceiling and the strong-vs-weak distinction reflexive, because every remaining chapter of Part VIII is measured against them.

The original papers (short, and worth it)

• Gene M. Amdahl, "Validity of the single processor approach to achieving large scale computing capabilities" (AFIPS Conference Proceedings, 1967). The three pages that named the ceiling. Amdahl's actual argument is a caution against over-investing in parallelism, and reading it in his own words fixes the law in memory better than any restatement. Tier 1.
• John L. Gustafson, "Reevaluating Amdahl's Law" (Communications of the ACM, 1988). The equally short rebuttal that reframed the question from "same problem, more cores" to "bigger problem, same time." The two papers together are §31.2. Tier 1.
• Herb Sutter, "The Free Lunch Is Over: A Fundamental Turn Toward Concurrency in Software" (Dr. Dobb's Journal, 2005). The essay that announced the end of automatic speedups to the software world. Written for C++ programmers, but the hardware story — Dennard scaling's end, the pivot to multicore — is exactly §31.1. Tier 1.

Modern Fortran, put to work in parallel

• Milan Curcic, Modern Fortran: Building Efficient Parallel Applications (Manning). The subtitle is the point: it carries the ideas of this chapter directly into coarray and parallel Fortran with running examples, and it is the natural companion for Chapters 32–35. Start here after this chapter. Tier 1.
• Metcalf, Reid, and Cohen, Modern Fortran Explained (Oxford University Press). The definitive reference for the language's native parallel model — coarrays, images, teams, collectives — which is the distinctive tool §31.3 previews. Reach for it when you want the exact standard semantics. Tier 1.

On the hardware and the laws

• John L. Hennessy and David A. Patterson, Computer Architecture: A Quantitative Approach (Morgan Kaufmann). The canonical text on why the hardware changed — power, the end of Dennard scaling, multicore, and Amdahl's Law applied throughout. The deep "why" behind §31.1. Tier 1.
• Alan H. Karp and Horace P. Flatt, "Measuring Parallel Processor Performance" (Communications of the ACM, 1990). The source of the experimentally-determined serial fraction used in Case Study 1 — a simple, durable diagnostic for real scaling data. Tier 2 (verify the exact citation).
• The TOP500 project (top500.org). The semiannual ranking of the world's fastest machines; browse it to see the scale — millions of cores — that makes weak scaling and Gustafson's Law matter in practice. Tier 1.

Free and online

• fortran-lang.org — the parallel-programming tutorials and the Discourse forum. Current, community- maintained introductions to do concurrent, coarrays, and OpenMP from Fortran, and the best place to ask a scaling question. Tier 1.
• The OpenMP and MPI specifications (openmp.org, mpi-forum.org). Primary sources for the two workhorses of §31.3; skim the introductions now for the models, and return for detail in Chapters 33–34. Tier 1.

Suggested order

1. Read Amdahl (1967) and Gustafson (1988) back to back — together they are barely six pages, and they are the intellectual core of this chapter.
2. Read Sutter's "Free Lunch" for the hardware turn, then skim the relevant Hennessy & Patterson sections for the quantitative "why."
3. Keep Curcic open as you enter Chapter 32; it is the bridge from these ideas to real parallel Fortran.
4. Bookmark TOP500 and the OpenMP/MPI intros; you will return to all three across Part VIII.