64-Bit Processors

[Vile]

Is it true Intel invented this technology but it had a bad response from the market, and AMD took advantage of that and managed to "make them famous"? Cause I always though AMD invented 64-bit technology in processors.

 

[paulbag]

I believe the Itanium cpu's were 64bit... but I don't know and I don't know if they were conceived before the Athlon64's came out.

 

[mdd1963]

"Cause I always though AMD invented 64-bit technology in processors."

AMD merely 'codeveloped' x86-64 coding with MS, a 64 bit OS while retaining the ability to run 32 bit code effectively...

Actual 64 bit OS's were in use before that already...Intels Floptanium ran special full 64 bit Winblows, various Linux variants, etc...

 

[slvr_phoenix]

Wow. Lots of bad information in this thread already.

Intel developed IA64 for Itanium long before AMD developed AMD64 / x86-64. Itanium's architecture however was nothing like x86. It was more oriented towards the server market that Intel was trying to break into than it was towards desktops.

AMD on the other hand, figured why reinvent the wheel. They just slapped 64-bit extensions onto the x86 that they knew and loved.

It's debatable whether or not this is true, but rumors of Intel developing their own 64-bit extensions to x86 existed before AMD said they were doing this. It's believed that Intel didn't openly persue x86-64 because obviously it would compete directly with the new server market they were trying to create with Itanium. Had Itanium done better, maybe they would have risked it.

But either way, as the concept of x86-64 caught on, Intel was forced to implement EM64T to follow on AMD64's heels lest they be caught without any product for that emerging market.

So it becomes debatable which between Intel and AMD actually pioneered x86-64. Chances are, both worked on different concepts of it at the same time, but AMD pushed it openly where as Intel kept it secret because AMD didn't have another 64-bit platform to protect, where as Intel did.

But either way, 64-bit processors themselves were hardly pioneered by either company, as they'd been used in servers and workstations for long long long before Intel and AMD ever bothered.

And x86-64 is most definately not an operating system. It's not software at all. It's just an instruction set extension allowing for 64-bit wide general purpose registers compatible with the x86 architecture. (And, of course, the hardware to run those instructions.) Operating Systems being software, are completely different (though related) animals.

 

[mpjesse]

If I'm not mistaken, DEC (Digital Equipment Corporation, now part of Compaq/HP) had the first 64 bit CPU. It was called the Alpha AXP.

See here:

http://en.wikipedia.org/wiki/DEC_Alpha

-mpjesse

 

[mpjesse]

Scratch that, MIPS had the first 64 bit CPU. Alpha was second. Though MIPS never caught on like DEC's Alpha did. See the timeline here:

http://en.wikipedia.org/wiki/64_bit

-mpjesse

 

[slvr_phoenix]

If I'm not mistaken, DEC (Digital Equipment Corporation, now part of Compaq/HP) had the first 64 bit CPU. It was called the Alpha AXP.

I was wondering about which came first. I figured it was probably Sun, but wasn't sure. I know DEC Alpha's have been around forever and a half though. (Feels like it anyway.) I've got an old Alpha box sitting under my desk with Red Hat and NT4. I don't think I've even so much as powered it on for years...

 

[slvr_phoenix]

Scratch that, MIPS had the first 64 bit CPU. Alpha was second. Though MIPS never caught on like DEC's Alpha did.

I'm not exactly sure that you can consider the Alpha as having "caught on"... It was a pretty big flop really, mostly for lack of software support as far as I witnessed. Every time I asked a software developer when they'd make an Alpha patch/port of their software, I was told "When Microsoft supports it." Never mind that I'd been running NT4, Alpha patched, and writing software in Visual Studio, Alpha patched, at the time of asking. :?

See the timeline here:

http://en.wikipedia.org/wiki/64_bit

Interesting. It's truly amazing the wealth of information contained at Wikipedia. Too bad a lot of it's junk and enough of it is wrong to make its dependability questionable. :lol:

 

[mpjesse]

Alpha may never have caught on in the desktop arena, but it sure as hell did in the corporate arena. (then again, alpha was never intended to be a desktop CPU... or even a workstation CPU).

Those bastards sold like hotcakes in clusters and mainframes. Mostly because of the high clock speed and awesome FPU.

Also remember that back then most large corporations didn't use M$'s then crappy server products. Everything was either done on Novell, UNIX (Linux to some degree), SunOS/SPARC's or IBM AS/RS systems. Both Novell and Linux supported Alpha very heavily. At the time, Alpha was a great alternative to Sun and IBM.

As for Wiki, I agree (for the hundredth time) that it's not a diffinitive source. However, it's better than almost anything else out there.

-mpjesse

 

[slvr_phoenix]

Mostly because of the high clock speed and awesome FPU.

Yeah, that was why we wanted to support Alpha as a workstation to run our software, since we write heavily computational scientific software and all. Unfortunately our software was already written for Windows when we went to support Alpha. Even then, we'd have done okay, except that to save time (and money) we purchased several third-party libraries. And while we could compile all of our code, no one was willing to compile theirs or allow us to do it for them. So only 95% of our software could be compiled for Alpha. That piddly 5% (the third party libraries) prevented us from ever releasing an Alpha build.

Which is why the Alpha box sits under my desk, not even booting up. We're using totally different software now, but no one cares about Alphas anymore.

 

[mpjesse]

Compaq killed Alpha. It turned out to be a poor investment on Compaq's part anyways.

But then again, no one ever counted on x86 ever being so successful in clusters, super clusters, and mainframes.

Today, 41% of the top 500 most powerful clusters are x86. EMT64/x86-64 is right behind with a combined 28%. PowerPC is up there at 14%.

Who would have thought?

-mpjesse

 

[mpjesse]

http://www.top500.org/stats

Slvr, run a search on "Operating System Family"

Look at Microsoft's share. LMAO. M$ still can't compete in the cluster market. They probably never will... there's not a whole lot of money in it.

-mpjesse

 

[mpjesse]

Now I'm on a roll...

Run a search for "June 1993" and "Processor Family"

Cray commanded a 38% top 500 share back then! Intel/x86 had just 13%!

At it's peak in June 1998, Alpha had 15.4% of the top 500 share. Alpha still has 4 superclusters on the top 500 list as of November '05.

-mpjesse

 

[slvr_phoenix]

Compaq killed Alpha. It turned out to be a poor investment on Compaq's part anyways.

The Visual Fortran acquisition from buying out DEC did them some good though.

But then again, no one ever counted on x86 ever being so successful in clusters, super clusters, and mainframes.

That's very true. Xeons were just too affordable to pass up though, and then Opterons forced people to rethink x86 as a 64-bit contender.

Today, 41% of the top 500 most powerful clusters are x86. EMT64/x86-64 is right behind with a combined 28%. PowerPC is up there at 14%.

Who would have thought?

Certainly not me. And definately not Intel. If they'd had thought it they never would have wasted so much money on Itanium. :lol: :lol: :lol:

Look at Microsoft's share. LMAO. M$ still can't compete in the cluster market. They probably never will... there's not a whole lot of money in it.

:lol: A Microsoft cluster. Imagine the cost. Imagine the downtime. He he he. No, even with their new cluster software, I doubt M$ ever will compete there. They just don't get it and there simply aren't enough people in the world dumb enough to give them a serious shot, not with so many better alternatives.

 

[mpjesse]

Certainly not me. And definately not Intel. If they'd had thought it they never would have wasted so much money on Itanium.

Truer words were never spoken. LOL.

But yeah ur right, the low cost of x86 Xeons was too hard to pass up. That's probably the only reason x86 has done so well in clusters. Generally speaking, one is not too concerned w/ how much space or power a supercluster uses. The only company that seems to really care about that is Google.

-mpjesse