from Hacker News

How MOS 6502 Illegal Opcodes Work (2008)

by soegaard on 4/20/25, 12:36 PM with 17 comments

  • by xp84 on 4/23/25, 4:46 PM

    I really enjoy how popular retro tech of the 6502 era has become (and thus how much content is created about the subject) from ~= 2017ish to now. For me it started on YouTube with 8-bit-guy and Adrian, but clearly it's in the zeitgeist in general, in a way that I don't feel like it was in the '00s or early '10s.

    Those of you who are Gen X and grew up with parents who bought you computers at a young age, you are so fortunate. Reading those manuals and learning to code in assembly gave that cohort an amazing fundamental understanding of computing. It's similar to the understanding of the Internet's underpinnings that we Millennial geeks gained by experimenting with HTML and the Web, before everything was sealed up and packaged for consumption.

    But I sure would love to experience some long summers as a kid in the early 80s with nothing but time and a Commodore 64 and its manual.

  • by msarnoff on 4/23/25, 2:38 PM

    Michael’s blog is one of my favorites. His “Ultimate Guide” talks are great and I rewatch them at least once a year.

    Ultimate Guide to the C64: https://youtu.be/ZsRRCnque2E?si=v4qQ9mCx3MafEwAq

    Ultimate Guide to the Game Boy: https://youtu.be/HyzD8pNlpwI?si=gn-F8yuRPJKoXmC7

    Ultimate Guide to the Apollo Guidance Computer: https://youtu.be/xx7Lfh5SKUQ?si=3qbvMohcGo6mBLrJ

  • by jmount on 4/23/25, 5:04 PM

    I grew up on 8088 and then 6502 and 8086 assembly. The 8086 had more registers and more useful variations of registers. However, it emotionally felt like after eliminating the accumulator and stack pointer: the 6502 had general purpose registers (though only 2 of them!) and that the 8086 did not have any general purpose registers. Even though there were more 8086 registers, each one had a different "purpose": base register, counter, data, base pointer, source index, destination index. These specializations never felt like a hardware optimization, but a series of horrific programming constraints.

  • by joelshep on 4/24/25, 5:54 AM

    This is great: thanks for sharing it. In 1983, Compute! magazine published yet another article on these opcodes: https://www.atarimagazines.com/compute/issue41/Extra_Instruc... . Now I can finally understand the why, not just the what.

  • by rzzzt on 4/23/25, 8:53 PM

    The technology (NMOS) is more important than the manufacturer (MOS). The CMOS variants from Rockwell and WDC irons these opcodes out to no-nonsense NOPs.

  • by monocasa on 4/23/25, 4:05 PM

    (2008)