3 ms·

> In 2003, AMD launched their 64-bit, but Pentium-compatible, Opteron CPU. Everyone stopped buying Intel CPUs for a while. [...] almost everyone immediately emb

> In 2003, AMD launched their 64-bit, but Pentium-compatible, Opteron CPU. Everyone stopped buying Intel CPUs for a while. [...] almost everyone immediately embraced AMD's instruction set and no one but HP wanted anything to do with Itanium. Huh, I never really connected the dots before, but TIL that this is why 64-bit images and package repositories of various Linux distributions were referred to as `amd64`...

krylon 5y ago
This leads to an interesting situation, where Intel owns x86 and licenses it to AMD, while AMD owns the amd64 extension and licenses it to Intel. Should one company revoke their license, so would the other (I assume, naively), and boom - no more amd64/x86_64 CPUs for you and me. Intel could probably create a new extension to x86, but that would take some time. And IIRC, AMD's x86 license evaporates automatically, if the company is acquired by someone else, which means the only company that could buy it without disastrous consequences is Intel, who probably would run into some anti-trust issues if they did.
- Beldin 5y ago
  What is it that ensures licensing is required? Patents? Because patents on x86 technology predating amd64 should be expired by now... and without patent protection, another company would be allowed to reverse-engineer it, no?
  - ramchip 5y ago
    https://www.blopeur.com/2020/04/08/Intel-x86-patent-never-ending.html https://www.blopeur.com/2020/04/08/Intel-x86-patent-never-en... Tl;dr: many new patents have been created over the years, for instance SSE extensions, and those haven't expired.
pjc50 5y ago
Yup. It was a classic mishandled transition; they messed up the emulation https://www.zdnet.com/article/intel-scraps-once-crucial-itanium-feature/ https://www.zdnet.com/article/intel-scraps-once-crucial-itan... , and getting the benefits of the transition required a big ecosystem shift. You more-or-less had to use Intel's compilers. Meanwhile AMD offered "x86, but faster and wider", which (once you booted a 64-bit operating system) could run either type of binary at native speed. Sometimes people really do just want a faster horse.
- _ph_ 5y ago
  It also didn't help that the Itanium project was delayed many years. Actually the first one "Merced", was supposed to come out 95 or 96. Had it been on time, things might have looked quite differently. And indeed, I think it was the AMD64 which breathed new life into x86, killing both the Itanium as well as most classical RISC architectures, because it was a nice upgrade path from existing PCs.
dekhn 5y ago
it was a great time. I remember buying a rackfull of AMDs at 1Ghz when the opteron first came out. They flew. They also consumed a ton of power and generated a ton of heat (I blew the circuit breaker for the whole floor of Evans Hall, in Berkeley, just turning them on).

krylon 5y ago

This leads to an interesting situation, where Intel owns x86 and licenses it to AMD, while AMD owns the amd64 extension and licenses it to Intel. Should one company revoke their license, so would the other (I assume, naively), and boom - no more amd64/x86_64 CPUs for you and me. Intel could probably create a new extension to x86, but that would take some time. And IIRC, AMD's x86 license evaporates automatically, if the company is acquired by someone else, which means the only company that could buy it without disastrous consequences is Intel, who probably would run into some anti-trust issues if they did.

Beldin 5y ago
What is it that ensures licensing is required? Patents? Because patents on x86 technology predating amd64 should be expired by now... and without patent protection, another company would be allowed to reverse-engineer it, no?
- ramchip 5y ago
  https://www.blopeur.com/2020/04/08/Intel-x86-patent-never-ending.html https://www.blopeur.com/2020/04/08/Intel-x86-patent-never-en... Tl;dr: many new patents have been created over the years, for instance SSE extensions, and those haven't expired.

pjc50 5y ago

Yup. It was a classic mishandled transition; they messed up the emulation https://www.zdnet.com/article/intel-scraps-once-crucial-itanium-feature/ https://www.zdnet.com/article/intel-scraps-once-crucial-itan... , and getting the benefits of the transition required a big ecosystem shift. You more-or-less had to use Intel's compilers. Meanwhile AMD offered "x86, but faster and wider", which (once you booted a 64-bit operating system) could run either type of binary at native speed. Sometimes people really do just want a faster horse.

_ph_ 5y ago
It also didn't help that the Itanium project was delayed many years. Actually the first one "Merced", was supposed to come out 95 or 96. Had it been on time, things might have looked quite differently. And indeed, I think it was the AMD64 which breathed new life into x86, killing both the Itanium as well as most classical RISC architectures, because it was a nice upgrade path from existing PCs.

dekhn 5y ago

it was a great time. I remember buying a rackfull of AMDs at 1Ghz when the opteron first came out. They flew. They also consumed a ton of power and generated a ton of heat (I blew the circuit breaker for the whole floor of Evans Hall, in Berkeley, just turning them on).