June 13, 2026
Carry chain? More like drama chain
The adder at the heart of Intel's 8087 floating-point chip
This tiny 1980 chip made math fly, and the comments turned into a nerd roast
TLDR: Intel’s old 8087 math chip got a spotlight for the tiny circuit that made its speed possible. Commenters loved the nerdy history, peppered the author with ultra-detailed questions, cracked goofy jokes, and complained that this famous chip still hasn’t gotten the modern remake treatment.
A deep dive into Intel’s 1980 8087 chip should have been a quiet history lesson about old computer parts. Instead, the comments turned it into a full-on retro tech fan convention with heckling. The big reveal from the article is that this famous math helper chip — the one that made certain calculations wildly faster for early PCs — relied on a surprisingly humble hero: a 69-bit adder, basically a tiny circuit doing the heavy lifting at the center of the action.
And the community? Absolutely locked in. The author jumped into the thread like a celebrity doing a post-show Q&A, declaring that adders and logic units are fascinating because every system does them differently. That invited the hardcore crowd to pile in with the kind of questions that make normal people slowly back out of the room: power delivery, missing metal layers, on-chip capacitance. In other words, the vibe was: “Cool article, now explain the microscopic plumbing.”
But the real scene-stealer was the joke comment mourning that there is “addition but no vipition,” which instantly gave the thread the energy of a classroom clown interrupting a physics lecture. Then came the spicy historical complaint: people have rebuilt Intel’s old main chips in modern code, but where’s the 8087 remake? That’s the closest thing this thread has to drama — not flame-war drama, but the classic internet flavor of polite disappointment mixed with impossible standards. In short: one tiny math circuit, one giant comment-section mood swing.
Key Points
- •Intel’s 8087 floating-point coprocessor, released in 1980, was designed to speed mathematical computation by up to 100 times.
- •The article identifies a 69-bit adder as the arithmetic core of the 8087 floating-point execution unit.
- •The 8087 die includes a Bus Interface Unit, a central microcode ROM, and a lower datapath split into exponent and fraction sections.
- •The adder is located in the fraction datapath, and exponent addition is performed by copying exponent data into that datapath.
- •To reduce carry delay, the 8087 divides addition into 4-bit blocks and uses a Manchester carry chain based on generate, propagate, and delete signals.