Slashdot Mirror
Intel Shifting 64-bit Plans
OS24Ever writes "News.com has an article stating that 'Intel plans to demonstrate a 64-bit revamp of its Xeon and Pentium processors in mid-February--an endorsement of a major rival's strategy and a troubling development for Intel's Itanium chip' Is this the end of Itanium?" Looks like the rumors were true.
Well we all saw this one coming with all the delays on the Itanium.
For "potential Itanium customers".
But based on their sales figures, it looks like they really aren't any.
If they had their heads in the right places, they'd heavily go after CT.
is solaris ported to this baby and theres 64 bit goodness for everyone!!
Until a 64-bit version of Windows comes out, I don't see this mattering all that much. 64-bit doesn't mean anything to the masses of end users, just the developers. I don't care if my computer is 2-bit or 1000-bit as long as it works well.
Intel has already publicly admitted to having X86 processors with 64 bit extension in development. Also, take a look at microsoft, who refer to X86-64 as "64 bit extended architecture."
Everybody and his brother figured out long ago that Itanium is not something that will penetrate effectively into the desktop market. It's hot, expensive, incompatible, etc. It requires a ton of work to get code running smoothly on Itanium. Th only amazing thing is how long it took intel to admit that it had egg on its face!
No, this does not signal that Itanium is doomed. Have a look at www.spec.org and look at the CPU2000 scores. Itanium is starting to kick some serious tail.
However Itanium is not a desktop chip-- its too big. 64-bit x86 will be a consumer product for desktops.
So the article doesn't really cover the issue I'm most curious about - are the x86-64 extensions (yamhill) compatable with AMD's Opteron or will they require different 64-bit binaries?
I love it when companies lie out their asses for months until they can flip their strategy around.
"Oh no, desktop users would never need 64 bit support! It's just not something a regular user ne-- CYKE! NOW HERE'S OUR LATEST AND GREATEST 64 BIT CHIP! PLEASE, NO CROWDING!"
It's been a long time.
The AMD64 veriant of of XP is still in beta, the only thing that seems to be holding it back is a 64-bit version of the .NET framework. Though I've been wondering lately if the delay is not the coding, but some more political, along the lines of "Intel and Microsfot are proud to announce the first 64-bit consumer desktop. Oh and *cough* support for AMD too".
The real question is have they finally dumped the stupid x86 instruction set in favour of a space/energy/coding efficient RISC set?
I thought we settled this back in the early 90s, there is no such thing as RISC versus CISC. The x86 is not CISC, the PPC is not RISC.
uhh, with all due respect, if we have learned anything in the past thirty years from the success of windows, unix, the as/400 and finally x86, it's that architectures are the hardest thing in the world to change due to the massive installed base, and that it's usually better to extend what you have.
just look at os/2, the MCA bus, and now itanium. why would i migrate to a new ISA and lose all the software that I already have when I can just grow my current one?
and x86 isn't that bloated, and cisc isn't that bad. just look at p4 vs. athlon - the tremendous clock speeds realized by the p4's use of an extended pipeline (which is a risc-like optimization) have a tremendous downside - you lose a lot of time resetting the cache if you miss a branch. so for interative programs, as opposed to massive number crunching (and that can be addressed cheaper using MPP and clustering), risc is something of a dog.
finally, you can't say that the desktop is not important to itanium when the line between servers, workstations, and desktops gets blurrier all the time, and the largest growing segment of the market is the low-to-mid-size server.
high-end servers may carry a premium price and have a higher margin, but like lenin said, quantity has a quality all its own.
this is not good news for intel.
>The real question is have they finally dumped the
>stupid x86 instruction set in favour of a
>space/energy/coding efficient RISC set?
Ok, yeah, right, umm....
You DO know that RISC processors generally take up a lot more memory space for a given program, have more instructions, and are often more complex to code for, right?
(of course this assumes you know what a delay slot is, or have understood the pain of manually doing indirect addressing, managing register windows during interrupts, or managing implicit instruction skip flags, the joys of RISC!)
I thought not..
as for the energy argument - get with the 90's - everyone is using similar internal execution units anyway - this is a red heering.
Of course, who am I to stand in the way of fashion..
RISC in it's pure form has not existed for over 10 years now.. neither has CISC, for that matter.
It's about the same as attacking russians for being communist.. it's just not that simple.
The x86 instruction set and successfully covered the widest range of CPU performance ever, and is available in by far the most computers... I would suggest by just about any measure it is by far the most successful ever.
Of course, there seems to be a group of people who cannot stand the pain of thinking about their python interpreter running x86 code internally, or the fact that gcc is generating that for them.
I truly feel sorry for them - they suffer on while the rest of us just get-on-with-the-job(tm).
Sigh.
all you had was a 286 @ 16 MHz to use
That would be one sad little lab. At the time the 286 was around, there were plenty of (dozens in fact) of scientific computing architectures vastly more advanced than the 286. They cost quite a bit more, too.
It wasn't really until the Pentium Pro came around that the processor architecture in 'mainstream' PC computing had caught up to the big boys. Since then, intel and AMD have largely been driving the cutting edge. This drove alot of them out of business, but even today there are niche markets who need serious I/O performance that intel machines don't deliver.
-
Actually, 64-bit computing isn't any faster than 32-bit computing. This is a common mistake made due to the surface facts.
In reality, 64-bit computing is possibly -slower- than it's 32-bit counterpart due to the increased bandwidth required, though smart engineering in modern 64-bit CPU's tend to work around this.
The advantage to 64-bit computing is, frankly, in the memory space that can be addressed. When you can address larger amounts of memory, you can make an application faster as less disk paging is necessary (assuming you have the memory to match). A good example of this are database servers. When you have 24GB of memory and a 20GB database, you can literally buffer the database in memory, this removing your slower disks from the equation.
Mind, you can do this with PAE on Intel's current 32-bit offerings, but I digress.
Ultimately, I think what Intel is -really- doing here is playing catch up on a modern variation of the "mhz myth game". Intel always took the hearts and minds of the average user, as a 3ghz P4 seemed better than an AMD processor running at 2.2ghz or a PowerPC running at 1.25ghz... even if in some or many cases, the "slower" chips worked faster.
Now, the average user is seeing the G5 at 2ghz, but a whopping 64-bits... and the Athlon64 chips at 2ghz, but a whopping 64-bits... and they're assuming that they must be faster due to their deeper bit depth. This is really nothing new. Sony has been doing this with the PlayStation2 for a few years now... claiming it to be a 128-bit system when it's really just a MIPS chip with a 128-bit vector unit. On this line of thinking the G4 and G5 are -also- 128-bit chips... but Apple just doesn't market them as such.
Intel had to act to counter this assumption, and the easiest way is to add 64-bit extensions to the P4, keep them clocked higher, and then win both of the wars.
Does the average user need 64-bit? No. Does the user who does know where to get it already? Yep. Sun, Apple, AMD, HP and even Intel's Itanium have been offering 64-bit technology for a while now.
This all comes down to marketing. That's it, that's all.
bash-3.00$ uname -a
SunOS panda 5.10 Generic sun4u sparc SUNW,Ultra-2
Whatever, man. I have G5 and Itanium2 machines at my desk. The HP Itanium2 runs Linux and WinXP 64-bit edition (which came out last June). The Itanium2 (McKinley) is an old slow one that crushes the G5 easliy on everything (using Intel's compiler) by factors of 2-3x. The new Madison Itaniums are substantially faster (look at the SPEC CPU benchmarks). The Itanium is far superior to anything else out there, it just doesn't run x86 code all that fast, and the GNU compiler sucks on the Itanium because the optimzier cannot get the VLIW right. The Itanium is just ahead of its time. And most people are too stuck in the x86 mindset to even see it. CPU buyers lose as a result.
Forgive me if I get a little bored by this 'revelation', I wrote about it in September:
:) )
http://www.theinquirer.net/?article=11668
And I followed it up a week later with this:
http://www.theinquirer.net/?article=11781
Ok, people seem to not taken me seriously then, so I'll reiterate. Prescott has 64 bit extensions built in. They use the AMD64 instruction set. This is because MS twisted their arm into it.
The question of when they turn it on is more a political one than a technical one, and that I don't know the answer to right now, most likely because Intel does not know either. They are in one hell of a bind. If Prescott is 64 bit, why should I pay 5x as much for an Itanic again? Oh yeah, a marginal performance gain on FP code, but a loss on Int. Whoopty-#&%^#-ding-dong.
It will be announced at IDF, count on that. When you can buy it, good question. My guess is that it will be an inticement for the first Prescott/EE buyers.
-Charlie
(As a self-plug, if you read the Inq, you would know these things
It doesn't matter if Itantium is better than AMD64, or Prescot64, or you name it -64. Alpha was better still, and it died. Itantium will die too because the other chips are good enough, and much cheaper. Intel will have to compete on price with AMD64, which makes Itantium a dead end.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
One company is just usually faster to the market with one new extension or another because they developed it themselves. If it takes hold, they either license it to the other or give it to the other under a current license. It's in Intel's best interest to keep AMD around to avoid being called a monopoly, so rather than let AMD die due to lack of a standard it needs to survive in the marketplace, they throw it a bone. (Think Microsoft and Apple) Of course, Intel makes it difficult by not licensing socket and slot types anymore, but the basic architecture is still licensed.
It wouldn't surprise me if Intel asked to see AMD's specs in case they wanted to one day include their technology & then used that info to build a compatable processor. I'm sure that Intel could use 3DNow! instructions if it wanted to, but simply chooses not to. (Why give AMD any credit for making something useful when you have other extensions that can do the same job?)
3DNow is somewhat of an extension of MMX, SSE was a response to that, and then you have 3DNow Pro and SSE2, etc. etc.. They just keep evolving the multimedia extensions. SSE2 seems to be the latest thing, so both Athlon 64 and Intel chips support it. I'm sure Intel will have SSE3 and SSE4 out soon and eventually AMD will license those as well --- if it doesn't already have rights to any new technologies from intel for the next few years.
More info on x86 extensions at Evolution of Extensions
Andy Glew (the designer of the Pentium Pro) on EPIC vs. normal architectures:
"Yes, but the IA-64 EPIC is not a modern architecture -
it is a design by committee, with microarchitects who believed
religious dogma instead of thinking.
At least some modern microarchitectures have made optimization
easier than in their predecessors. Apart from some egregious
glass jaws (mea culpa), P6 was often less sensitive to optimization
than the P5. The compiler folks complained that their unoptimized
code often ran as fast as their optimized code.
AMD's K7 and K8 continue in this vein.
This is one of the reasons I jumped from Intel to AMD:
the Intel P6 is philosophically a lot closer to the AMD K7 and K8
than it is to the Intel Pentium 4 (Willamette, Prescott), or Itanium.
Pentium 4 is fragile, just like Itanium."
Look at how many trouble Intel has trouble admitting they are wrong and following AMD on something. Intel is so used to AMD copycating them on everything, they don't seem to know how to deal with the shoe being on the other foot.
So far Intel has followed AMD onto DDR memory, after dragging their feet for a year. Now it's happening with 64 bits. Next expect to see it with integrated memory controller, desktop dynamic power management(like quick 'n cool) and hypertransport. I'm sure when they come around the technologies might be similar, but they'll have some other name for it. Hopefully, Intel doesn't try the old Microsoft embrace and extend.
Will Intel's extensions to the x86 architecture be compatible with AMD's. Or will fat binaries that can execute in x86, x86-64 Intel and x86-64 AMD be necessary?
cheap labor conservatives - they want to keep you hungry enough to be thankful for minimum wage.
You are wrong on both.
Windows 2000 was released as a release candidate for Alpha and it was a real 64-bit Windows with 64-bit pointers just like the current Itanium versions of Windows.
Stick six megs of cache on a Xeon, and if the Itanium still wins by a substantial margin, then I'll think that Intel hasn't wasted the last ten years and billions of dollars.
Nothing wrong with sticking a lot of cache on a part -- everyone would, were it not for other issues such as cost -- but that Itanium is better than anything else does not follow.
Itanium puts up impessive numbers, that I can't deny. I'd expect any competent architecture with that much raw die area thrown at the problem to do the same, though. There's little indication that any of the performance gains are due to the architecture of Itanium. In fact, there was an ISCA (?) paper by Intel which reported that major features of Itanium -- eg branch elimination through predication -- were worth a little if you hand-tuned, nil if you had a decent (intel) compiler, and negative if you didn't (gcc at the time).
Which is all just a way of saying that Itanium is just another architecture. It tried some things that worked, some things that didn't, and in the end does well because the ones making it can throw tons of resources at the problem. "Ahead of its time"? No, because in the future, the same thing will be true.
The enemies of Democracy are