Hacker News Reader

Comment by johnklos | Hacker News Reader

This is from a school? The author (Doeppner, it seems) should be embarrassed.

"x64" is a marketing name primarily used by Microsoft. Sure, many of us know what you're talking about when you use "x64", but it's just not a real thing. There's no 464, 564, 664, 764, et cetera, so the "x" means nothing, unlike its use in "x86".

It's like using "virus" to describe a Trojan - people will generally get the gist, but you're technically wrong, and if you're supposed to know what you're talking about, then you're certainly not showing it.

View on HN

Yeah, it's actually called EM64T or at the very worst IA-32e. /s

Speaking seriously, there are two variants of this architecture: AMD64 and Intel 64 (that's what AMD and Intel call their respective versions). Everything else is misnomer, including legacy monikers such as "x86-64".

As a recent grad, I was going to say education on hardware optimization is lacking in universities, but then again there's a lot of snake oil on sites like stackoverflow. Although users like Peter Cordes have done a good job cleaning up, people still seem to eschew benchmarking for eyeballing code à la "Big O", ignoring the effects of compilers and hardware (CPU type, memory layout).

I wouldn't be too harsh on the prof for a throwaway doc: it reads as a perfectly fine intro to concepts like stack, heap, and basic assembly instructions. I'd argue most computer science students, with aims of a programming job, don't need to even know these concepts [1], [2]. With hardware advancements, most probably don't need to worry about code performance. Of those who do, they'll just google Agner Fog, and learn on their own just fine. Despite my opening sentence, I'm sure the Electrical Engineering/Computer Engineering departments are much stronger in this area.

[1] - https://en.wikipedia.org/wiki/Global_interpreter_lock

[2] - https://en.wikipedia.org/wiki/JavaScript

This is one of the most stark examples of bikeshedding I've seen on this site.

> one of the most stark examples of bikeshedding

I think your reaction is an excellent example of a significant problem with discourse on the Internet.

Correct terms matter. Using incorrect terminology just because it's common is no excuse.

Being pedantic about things isn't helpful, either, but which is worse: a teacher in a specific field using terms incorrectly, and therefore passing along that incorrect usage, or someone pointing that out?

So let's reexamine: is this "one of the most stark examples of bikeshedding", or is this the Internet showing its lack of tolerance for a genuine care for facts and correctness?

So, what is the correct term then: "Intel 64" or "AMD64"? That's what the CPU manufacturers call their specification of that architecture right now. Intel used to call it "IA-32e". AMD called it "x86-64" before they've released it, but then switched to "AMD64". All of those are "marketing names" used (or no longer used) by the manufacturers. How should we pick the correct one? Is there even a correct one?

If "x64" can refer to amd64 or Intel 64, then it can also refer to PPC64, aarch64, mips64, et cetera.

But Intel 64, amd64, x86_64, x86, all work because none are ambiguous or misleading. "IA-32e" would be confusing, but at least it refers to a specific Intel marketing term that one can look up. They're specific enough to avoid ambiguity.

(to be fair, "Intel 64" could refer to Itanic, so that's not as unambiguous as it could be, but without context, it's reasonable most would assume it refers to 64 bit x86)

> If "x64" can refer to amd64 or Intel 64, then it can also refer to PPC64, aarch64, mips64, et cetera.

No, it cannot. It refers exactly to AMD64 (not "amd64", it's a GCC's marketing name) and Intel 64 (and their previous revisions) and nothing else. See, I too can omit middle terms in a logical sequence and simply assert the conclusion (You've omitted something like "and if the name can denote two things it may as well denote any number of things with vaguely similar alternative names" which is obviously ridiculous).

> "Intel 64" could refer to Itanic

By your own logic it could not. "Itanium" is a marketing name, the architecture's actual name was "IA-64".

The things have names that people put on them. If people refer to AMD64 as "x64", then "x64" is a name of that thing, literally. Whether it's correct or not is a matter of convention and popularity, and your proposed convention is quite unpopular, I'd say.

Ah. The logic of the masses. Apologies that this was a bit deep for you :)

I vote bikeshedding

Sorry but you veered way past "someone pointing it out" and crashed your car into the bike shed when you said "The author (Doeppner, it seems) should be embarrassed."

Imagine having studied tons of literature for your thesis and then ending up in a industry that uses the same terms, except with half of them with a meaning that is incompatible with what you learned, but in a subtle manner.

and then there's x32, the (abandoned?) project for taking advantage of most of AMD64 while keeping pointers as 32bit to conserve memory

Reply on news.ycombinator.com