Fading out audio is one of the most CPU-intensive tasks you can possibly do!

When numbers get *really* small, the number of CPU operations can explode 100-fold.  Both x86 and ARM have special CPU instructions just to handle it.  But why?

In this video, we'll explore the IEEE 754 (floating point) standard, the fight between Intel and DEC, and I'll write some demonstration C++ code that illustrates this problem even today!

---

Timestamps:

00:00 Subnormal Arithmetic Cost
02:25 An Accuracy Debate...
06:28 Too small to calculate?
08:56 IEEE 754 Standard
10:17 Digital Audio Workstation Conundrum
14:18 A Massive CPU Spike

---

Follow LaurieWired on Social Media:
►https://linktr.ee/lauriewired

---

Papers Cited:

"An Interview with the Old Man of Floating-Point":
https://people.eecs.berkeley.edu/~wkahan/ieee754status/754story.html

Pentium 4 and Xeon Processor Optimization manual  (Chapter 5, “Optimizing for SIMD Floating-point"):
https://kib.kiev.ua/x86docs/Intel/Intel-OptimGuide/248966-004.pdf

---

Short demo code:
https://gist.github.com/LaurieWired/3f8be4648b3e627e3fcd5607d83b405f

Laurie Wired's resource offers insights, tutorials, and resources for technology enthusiasts, DIYers, and makers. Readers can learn about electronics projects, hardware hacking, and DIY gadgets. With articles, tutorials, and project guides, Laurie Wired provides resources for learning and experimenting with technology.

LaurieWired

Audio fading operations can cause severe CPU performance degradation due to subnormal floating-point numbers. When audio volume is gradually reduced, calculations slip into the subnormal range, which can cost hundreds of additional clock cycles per operation. This issue affected major DAWs in the early 2000s with Intel's Pentium 4. The problem stems from the IEEE 754 floating-point standard, which introduced gradual underflow through subnormals for arithmetic precision. While this provides mathematical accuracy, it creates significant computational overhead. Modern audio software typically disables subnormal calculations by default, flushing tiny numbers to zero to maintain performance without perceptible audio quality loss.

Fading Audio is ROUGH on CPUs