Intel 64-bit Announcements at IDF 518
cribb writes "Some fascinating stuff is going on over at the IDF. Ever since the first sneak previews of the Opteron, there has been lots of uncertainty around its future, and that of AMD. AMD have bet everything on the success of their new 64-bit CPU, and with Microsoft severely delaying the release of a 64-bit Windows, and Intel complaining that 64-bit processing has no place in the desktop market, things were starting to look dim for AMD. However, after rumours around the 64-bit extensions of the Pentium 4 EE, it became clear that Intel is not willing to lag behind AMD in the 'innovation' department. Now comes the shocker: Intel boss Craig Barrett today anounced that Xeon-class 64-bit server CPUs codenamed Nocona will be coming out the second half of 2004. It isn't clear whether they will support AMD's Opteron AMD64 extensions. Barrett is quoted saying, 'There will be one operating system that will support all (64-bit) extended systems.' Maybe 64-bit computing is right around the corner after all, and we may even see compatible instruction sets from Intel and AMD! And does this mean that Intel will be dumping Itanium, which never caught on as expected in the server market, and forget the billions spent on developing it?" See some other articles at EE Times, and EWeek.
I think intel has only a small storm to weather (Score:3, Informative)
Cheers for AMD and their success wit x86-64.
Completion is best for everyone in this game.
Re:Quote (Score:3, Informative)
Windows 64-bit preview available (Score:5, Informative)
Intel's new 64-bit CPU's (Score:5, Informative)
Inquirer.net (Score:5, Informative)
Essentially there will be a single OS for the two (Intel and AMD). Unspoken is that Intel's implementation is AMD64 ISA, but a different technical architecture. If it's compatible, who cares. Secondary confirmation via Ars Technica [arstechnica.com]
And lest we forget... (Score:4, Informative)
More info (Score:5, Informative)
News.com article [com.com]
Intel's 64 bit extensions are compatible with AMD's. You will be able to run the same 64 bit OSes on them. Intel's 64-bit capable Xeons are Noconas, which are Prescotts in a Xeon package.
I work for Intel, but I do not speak for Intel. My opinions are not necessarily the opinions of Intel Corporation.
The Register (Score:4, Informative)
Intel won't say if it has licensed AMD's x86-64 extensions. But Barrett seemed to hint that Intel's technology will be somewhat less than completely compatible with AMD's instruction set.
"For the most part, (software) will run on both systems," he said. "Intel has some (things) unique to Intel, which we will make sure people write, port and tune to."
--
Sigs are for geeks
Missing step (Score:5, Informative)
Intel's shortcommings in the IA32 to IA64 switch were the following:
In the meantime, AMD took the evolutionary path and provided the 64-bit capability from desktops, to middle tier servers and higher end machines. They implemented an architecture that directly executes the IA32 but that was extended to the much needed now 64-bits. The performance / price ratio are much better than that of Itanium's and compilers were much easier to come about since the x86 ISA is a well known one.
There is no surprise that AMD made the right strategic move to provide the needed missing link in the evolution of the popular (but crappy) x86 ISA to the 64-bit arena. There is no surprise either that heavy weights such as IBM, Dell, SUN and even HP -- who pretty much designed Itanium -- put some of their eggs in their AMD busket.
And there is no surprise that Intel realized after the fact that it should had provided the missing step and it is now playing catch up.
Isn't unbridled competion good? The pervasiveness of Intel forced the AMD and the RISC designers to do their best to improve their own designs which now in turn are forcing Intel to improve its own?
The same story with UNIX/Linux and MS windows.
People need decent alternatives to chose from. Forced monolithic single-vendor solutions are bad for everyone.
AMD Low Power (Score:5, Informative)
More information: AMD [xbitlabs.com], Intel [xbitlabs.com] at xbit
Discussion: AMD [aceshardware.com], Intel [aceshardware.com] at Ace's
Re:Intellectual Exercise (Score:5, Informative)
The iAPX 432 [brouhaha.com] was a 32-bit processor Intel developed starting in 1975 that embodied CISC technology to the max. It was innovative, but also expensive and slow, and targeted towards the Ada programming language, another market failure.
Researchers in University. (Score:5, Informative)
Re:64 bit screw up (Score:3, Informative)
Second, int is 32 bits on most 64-bit platforms (PPC64, SPARC64, etc.).
Third, long is the same size as void* on virtually all modern platforms, so that's the assumption people should be making.
Link to Intel 64 Bit Extensions (Score:5, Informative)
From the Intel FAQ Site:
Q9: Is it possible to write software that will run on Intel's
processors with 64-bit extension technology, and AMD's 64-bit capable
processors?
A9: With both companies designing entirely different architectures, the
question is whether the operating system and software ported to each
processor will run on the other processor, and the answer is yes in
most cases. However, Intel processors support additional features, like
the SSE3 instructions and Hyper-Threading Technology, which are not
supported on non-Intel platforms. As such, we believe developers will
achieve maximum performance and stability by designing specifically for
Intel architectures and by taking advantage of Intel's breadth of
software tools and enabling services.
HP is gonna be pissed.... (Score:1, Informative)
Like a Crystal Apple (Score:2, Informative)
1. 64 bit computing [pcworld.com]
2. Bluetooth [wirelessdevnet.com]
3. Firewire [com.com]
4. 802.11b/g [com.com]
5. USB [techtv.com]
6. DVD/CD Writeable [got tired of linking]
100,000,000. SCSI
Re:More brainless ad campaigns... (Score:4, Informative)
Re:Why 64 bit? (Score:5, Informative)
2^64 is 18446744073709551616. This is BIG. 17179869184 gigabytes. 16777216 terabytes of addressable memory. 16384 petabytes. This is basically the maximum amount of physical memory and the maximum size of one individual process's virtual memory mapping on a 64 bit architecture (yes, I know many current 64 bit implementations, including AMD64 are limited to 2^48 in practice; but the architectures can fundamentally handle both 2^64 physical and virtual addressing).
This is enough addressing that you can have 2.5GB of memory in a process for each man, woman, and child on the face of the planet.
And as to doing integer math larger than 2^64-- why? 2^32 is already overkill for most things.
Nope, I don't see "128-bit computing" becoming mainstream anytime soon. And it's far from clear 64 bit on the desktop is all that close, given the fact that A) the added code size contributes cache misses and saps performance, and B) there is not much done on the desktop now that requires more than 2^32 bytes of memory in a process, and C) not much stuff does math on quantities greater than 2^32 (4294967296). Keep in mind bank switching allows you to have more RAM than 4GB on all recent ia32 processors (2^36/2^40).
If we change architectures, it will be less about addressing limitations and more about the piss-poor quantity of registers available on ia32. More registers means more obtainable instruction-level parallelism.. this equals more work done on modern architectures.
correcting myself. OSDL not OSDN. (Score:2, Informative)
Intel helped start (and Linus is an employee of,) OSDL [osdl.org]. That one letter makes a heck of a difference.
Re:Intel wouldn't ditch Itanium... (Score:5, Informative)
The other major RISC CPUs sell by the millions. Your whole post is one big pointless troll.
128-bit computing? (Score:5, Informative)
Well, the designers of IBM's venerable AS/400 might disagree with you. Its architecture has been enabled for 128-bit computing since the early 1970's [channelweb.com].
Re:The Future... (Score:4, Informative)
This how it typically works in UNIX-land: 64-bit kernel with 32-bit and 64-bit applications. Solaris has been like this for three generations, IIRC. For general day-to-day use, it is typically 99.9% 32-bit applications and the occasional life-saving 64-bit app for that immense data file or whatever that comes up.
Re:64-bit misinformation rampant in the press (Score:3, Informative)
How the hell did the parent get modded up?
Re:Will not sink the Itanic (Score:3, Informative)
Theyre shipping, theyre just real expensive
I'm still bitter about the Athlon MP - which was supposed to allow for 4-way Athlons.
IANA chipset designer, but AFAIK, that's a physical impossibility, there's only one CPU select pin on the socket A, which'd allow for merely 2 cpus.
Re:Intel wouldn't ditch Itanium... (Score:5, Informative)
The G5 is a PowerPC. PowerPC is RISC (or was, due to the blurring over the years). Apple is putting G5s into their servers, too.
As far as numbers go, the market reports I see via Google put Sun's annual server sales at 200,000 to 300,000 servers. Multiply by some SMP co-efficient, and it isn't hard to get over 1,000,000. These numbers also do not include workstations, motherboards sold to OEMs, embedded units, and replacement parts. If Itanic is just getting over 100,000, then Intel has some problems, espcially considering just how long that CPU has been around (I first heard about it in college in 1997 or so, I think).
possible answer to question of compatibility (Score:2, Informative)
Re:Quote (Score:3, Informative)
Intel was also a major early investor in Red Hat...
Re:Is this a surprise to anyone? (Score:2, Informative)
There are even now a few things that 64 bit is useful for...
1. Crypto - 64 bit ints makes things much faster.
2. Running Lisp or Scheme. You can fit a lot more into a 64 bit int, which is important when running these languages.
3. Some calculations can be done in 64 bit ints that would not fit in 32 bit. Example, financial calculations, where $3,000,000.00 does not fit in 32 bit.
Itanium was designed to emulate x86 (Score:2, Informative)
But you have to wonder... what on earth was Intel thinking?
It wasn't x86 compatible by a longshot
Intel was thinking that Itanium would be enough faster than a native x86 to emulate x86 competively. The architecture was designed to make software emulation of x86 relatively efficient.
The trouble is, the required performance never arrived. Clock rates greatly lagged x86. Compilers have not been able to use the resources provided by IA64 effectively. Itanium hasn't been able to keep up with x86 when running native, much less in emulation.
Re:Intel wouldn't ditch Itanium... (Score:4, Informative)
So Power4 is the only processor you'll admit to being a "like" processor? The biggest RISC processors are Sparc and PowerPC. Each out sells Itanium by a huge margin.
Sun sold more total systems, but they weren't all SunFires, and we weren't talking about them.
Intel sold 100,000 processors, Sun sold close to 300,000 systems. See the significance of the difference? The parent of this thread was talking about RISC processors. Power4 was mentioned specifically, but Sparcs are certainly in the the same class.
Re:Why 64 bit? (Score:3, Informative)
Maybe you should check out some benchmarks on something newer than a 486, because most current CPUs can actually do floating point calculations faster than integer ones.
Re:Missing step (Score:4, Informative)
But you can excuse DEC for the mistake - they had migrated their customers to a radically different architecture before. When DEC killed the PDP-10 line, they put in a lot of effort to move their existing customer base for that machine to the VAX. And it worked. People may have bitched, but they moved because there was no real alternative then (the only other 36-bit line at the time was Univac, and they were getting ready to throw in the towel). So the DEC customer choice was either another DEC machine or an IBM mainframe of some sort. And guess which one their customers chose? Sure they bitched about it, but it wasn't as if they had any real choice in staying with some sort of comaptible system. And most of their software that wasn't written in MACRO-10 or Bliss was tied to DEC Fortran or COBOL.
It's clear that when DEC did the switch to the Alpha, they expected something similar to happen. The few things they didn't notice? First, there were other 32- and 64-bit platforms to migrate to. A lot of the customers took the opportunity to look at SPARC or MIPS or (GASP!) Intel 32-bit offerrings as well as the 64-bit goodness soon to come out from the other two. Second, most customer's software was not as tied as heavily to their platform. In the interrim, code had migrated to C, FORTRANs and COBOLs had become much more standardized, and very few folks wrote in MACRO-32. Toss in the fact that it's a lot easier to port a program from one 32-bit platform to another 32-bit platform and it's no wonder that DEC's customer base ran away screaming. And that was the end of DEC.
Now Intel, OTOH, has gone through this with at least two other architectures - the IA-432 and the 9900(??) - you'd have thought they's learned their lesson by now. Oh well, third time's a charm - maybe thry'll introduce the 128-bit extensions next year to retake the lead!
Not for nothing, but... (Score:2, Informative)
I think Microsoft was having similar trouble trying to adapt Windows to run properly and without issue on this brand new hardware. Maybe AMD was dragging their feet addressing errata exposed by the effort?
But even then I feel that's a bit of a stretch. It's been in beta state for a long time now. At least they're offering security updates for it... that's a sign of commitment.
The G5 is NOT a 64-bit "machine" (Score:3, Informative)
If your definition of 64-bit is a 32-bit operating system around a 64-bit chip, then the G5 is a 64-bit platform. Mac OS X 10.2.7 (and the upcoming 10.3) is not a 64-bit operating system [osnews.com]. This is particularly frustrating because Apple's marketing machine has very carefully crafted their message [apple.com] to make a reasonable person believe the operating system is 64-bit, especially if you download and read Power Mac G5 Tech Overview [akamai.net] (PDF). Apple says about the G5 version of Mac OS X that it runs all of your software -- and runs it faster -- with a version of Mac OS X Jaguar specially tuned for the PowerPC G5 processor, providing a seamless transition to 64-bit power. That's only the beginning of the smoke and mirrors. The 64-bit power only gives users two things: the operating system can address up to 8GB of RAM, though user programs are still limited to 4GB, and some of the G5 numerical hardware is available with a special version of GCC (3.3).
On the iAPX-432 and the Ada programming language (Score:4, Informative)
First, your statements above are contradictory, in 1975 there was no Ada programming language, only a spec (steelman ??) that described what the language should contain (and not contain).
Also, it is not clear whether you meant that the 432 or Ada was a marketing failure (or both). Certainly the 432 was. OTOH, from its first release in 1980 or so, the Ada language has been far from a "market failure", despite there being no low-cost compilers for it and despite the limitations required by the SteelMan spec. Virtually all aeronautics, astronautics or critical communications software (Military or civilian) and weapons control software for the last 20 years was written in Ada (and not just in the US).
In addition, several commercial SW firms also found, even w/ Ada-83, that it allowed them to ship w/ far fewer bugs left for customers to find that code written in (Ugh!) C, as well as allowing bug-fixes using less than 50% of the developer resources than to fix bugs in (Ugh!) C.
As of 1995 the Ada language is much more oriented towards general programming, as well as being much cheaper to use than it had been. There has been a FREE (GPL) Ada compiler available since 1995 or so, and it is now (since version 3.2) integrated into GCC.
For more info on how Ada is being used and why it should be used for all new projects, see My small Ada site [nyct.net] or David Botton's Ada Power [adapower.com] site.Re:i960 is not like the i432 or Itannic (Score:1, Informative)
Re:I'm not surprised at all. (Score:3, Informative)
While I think the Opteron is a great choice, and kudos to your company for getting the best of what they can right now, many companies are also either blindly loyal to Intel or don't want to adopt anything too "new" too quickly. The fact that a few linux distros and now some beta software from MS will take full advantage of the Opteron will probably have an effect too, despite the Opteron running 32-bit code very nicely. It's the uninformed managers that tend to make those decisions.
Again, congrats to your company which decided to jump on already. I'm still of the opinion that there will be plenty of 64-bit sales for Intel and this is no more than a minor scrape, completely detached from a favor for either company.
Re:See the doc: IA32-e is in fact x86-64 (Score:2, Informative)
For AMD see the table 1.1 "Operating modes" page 43 of http://www.amd.com/us-en/assets/content_type/whit
For Intel see the table 1.1 "IA32-e modes" page 18 of http://developer.intel.com/technology/64bitextens
(I dont know why slashdot add sometimes a space in the URL, it's not in the original)
The CMPXCHG16B example just show that Intel continue there nasty game of adding opcode in a way that nobody can use it because thre need to run on AND chip too.
Re:i960 is not like the i432 or Itannic (Score:3, Informative)
Re:On the iAPX-432 and the Ada programming languag (Score:1, Informative)
I know that at least a few Australian universities teach Ada first.