It's official now: http://news.com.com/2100-1006_3-5160169.html?tag=nefd_top

Intel will add 64-bit extensions to their Xeon, and later their P4 CPUs.
They will be compatible with AMD's 64-bit extensions.

Bad news: we might be stuck with x86 for another decade or so.
Good news: we won't be stuck with AMD, but we can also have Intel.
Posted on 2004-02-17 16:08:47 by Henk-Jan
Am I the only one reading it like: Intel will be compatible with AMD64 but not the other way around?
Posted on 2004-02-17 16:42:43 by C0D1F1ED
I have 1 question which is quite off topic, it's been bugging me for years now and still haven't found a satisfying answer:

Why didn't Intel implement AMD 3DNow instructions(or AMD specific instructions) on their chips? if AMD can implement SSE/SSE2... why can't Intel do the other way around? Are there legal problems in doing so?

I'm no Intel fan nor AMD.
Posted on 2004-02-17 18:45:34 by arkane
Well I am an AMD fan so this is quite a happy moment for me, but this is quite a shock for me as I though Intel was going to go with enhancing the IA-64 arch and putting it in its line of chips. Looks like Intel is following the so-called "underdog" of PC microprocessors.
Posted on 2004-02-17 18:49:39 by x86asm
Am I the only one reading it like: Intel will be compatible with AMD64 but not the other way around?


Could be... Does it matter though? AMD has been not-entirely-compatible for years, lagging behind with MMX, SSE, SSE2, and now SSE3.
You can always program for the lowest common denominator.

Why didn't Intel implement AMD 3DNow instructions(or AMD specific instructions) on their chips? if AMD can implement SSE/SSE2... why can't Intel do the other way around? Are there legal problems in doing so?


Are there any technical reasons why Intel should implement it? 3DNow! is just a silly hack on MMX registers. SSE is vastly superior, no shared registers, registers twice as large... I'd much rather have SSE than 3DNow!. Else you might aswell ask why they don't have a Z80 mode implemented, or something. There are no legal problems by the way. One of the clauses in the x86 license is that Intel may freely use any extensions that any licensee may add to their x86-clones. This just seems to be the first occassion where Intel actually makes use of it.

I though Intel was going to go with enhancing the IA-64 arch and putting it in its line of chips. Looks like Intel is following the so-called "underdog" of PC microprocessors.


Intel has had both IA-32 and IA-64 product lines for years, I see no reason for them to change this, as the IA-64 is probably still a lot faster when it comes to the heavy stuff. After all, it's a much 'larger' CPU... More functional units, more registers, more cache...
They haven't said anything about abandoning IA-64 yet, so I don't think they will.
As for following the underdog... We'll see who will be following who when the first performance-benchmarks arise :)
Posted on 2004-02-17 19:29:22 by Henk-Jan
Be careful how you code because the competition between AMD and Intel serves us well! It is wrong to exclude one or the other. I optimize for AMD because I can only keep up with one company's wares, but it will always run on Intel.
Posted on 2004-02-17 21:17:49 by bitRAKE
Duh.

Why is anyone surprised? Intel and AMD have had a tech sharing deal for a long time now.

Intel probably knew the new instruction set before anyone else.
Posted on 2004-02-17 23:57:44 by ThoughtCriminal