“Intel will renovate its complete desktop processor product line in 2006. 65 nm processors will replace current single and dual-core Pentiums, Yonah will be officially introduced as mobile and desktop processor for small form factor devices. But Intel will not be able to increase speed levels dramatically until the arrival of its next-generation processor architecture, Tom’s Hardware Guide has learned,” Wolfgang Gruener reports for Tom’s Hardware Guide.
“With the transition from 90 nm to 65 nm processors scheduled for the first quarter of 2006, the company plans to launch five new product families just for the desktop segment: The 65 nm “Cedar Mill” will replace the current 90 nm Celeron D and Pentium 4, the 65 nm “Presler”, to be named Pentium D 900, will replace the current 90 nm Pentium D 800 (Smithfield core), the 65 nm “Conroe” is designed to replace the Pentium D 900 late in 2006, and the next-generation Pentium M “Yonah” will be a completely new product for the desktop market,” Gruener reports.
Full article here.
Zupchuck:
I agree with your comments on D.I.I., but I have a question: What makes you think Intel is using Apple to shed it’s Wintel baggage? After all, it’s being billed as a move to x86 – that’s even the checkbox on the Xcode compiler – and x86 and Itanium, or x86 and any other CPU architecture, are mutually exclusive.
So, they’re either staying with x86 or they’re not. No?
But just focusing on the article, I guess I’m the only one a little underwhelmed by this announcement. It’s all the same clockspeeds using all the same basic designs (for each version), just on a smaller chip, and with the promise of something ‘really’ new in late 2006 (more likely early 2007 at best). If Apple’s the ‘Intended One’ for these new history-free designs, it’s gonna be a looong wait.
65nm is great, but it really doesn’t seem to be improving Intel’s product much – just improving their yields. No real performance boost for the consumer, while they get to make a few more bucks for the same price … I’d say that’s about typical for Intel.
Odyssey67,
Is the checkbox “Intel” or “x86”? Does Apple depend on the legacy part of x86 or not?… Windows requires it. Apple does not.
The move is being billed as a move to Intel. Steve Jobs makes no mention of x86.
I used to know a guy whose name is Yonah. He’s Jewish.
Intel’s main R&D operation is in Haifa, Israel, so don’t be surprised
if Intel uses Jewish names for their processors.
Wait and see. Can we believe Intel?
Zupchuck:
My eyebrows were raised just as much as everyone’s, upon seeing the WWDC show’s graphics simply indicating “Intel”. However, while I admit that because its been over a month my memory may be fuzzy, a WWDC screenshot I saw said something like “Intel x86”. Besides, neither Apple or Intel has made any indication that a whole new, or Itanium based, CPU is what’s in store. Also, they said Windows would run on whatever they wind up selling, and Windows will only run on x86 (without something like VirtualPC that is). Finally, unless the CPUs in the developer’s rigs were intentionally mislabeled, System Info said it was a 3.6Ghz Pentium 4, which is x86. So, I think it’s safe to assume.
The x86 instruction set – any instruction set – is integral to the CPU. So, for it to work, OSX would have to be just as wedded to x86 as any build of Windows. That’s why they will include Rossetta for the PPC applications – Rosetta imitates in software the hardware based instruction set found on PPC CPUs, so that the applications will work on x86 CPUs. I imagine that’s also why their will be no Alitec support in Rossetta; Altivec ‘hardware’ is providing performance enhancements that simply can’t be replicated in a software emulator/translator. This is also another clue that we’re dealing with x86. If Intel was building Apple a new CPU, with somekind of new architecture, it would probably be smarter – or at least more developer, and installed user base, friendly – to at least include some version of Altivec on it, rather than go with a non-Altivec optimized Rosetta solution.
odyssey67,
One does not have to use the whole instruction set. Microsoft is wedded to ancient parts of it, Apple is not. Here’s an interesting link: http://psychcentral.com/psypsych/X86
So what that “Windows will run on whatever [Apple] winds up selling.” Windows runs on PPC, too – through VPC. Steve is famously ambiguous and lets your imagination connect the dots the way your perception drives it.
OSX is not wedded to x86 as any build of Windows – it runs on PPC and Intel (currently P4) processors. Windows IS (but then Microsoft does build versions of Windows for other processors, too).
What is to stop Apple from using the RISC-like instructions directly on a CPU and not using the x86 translation also built on the chips? It’s this translation that is burdensome.
Apple has already transcended the CPU dependency. The fact that OS X is no more dependent on PPC than x86 Intel means Apple isn’t dependent specifically to the whole x86 instruction set.
I’ve said this before when rumors first started popping up just before the Intel announcement. Apple will use new chips from Intel that will blow all the competition out of the water. Intel is currently prepping a new chip that will leave IBM’s Cell processor in the dust when it comes to multicore processing and whatever else the rest of the competition is touting right now. Steve Jobs saw IBM’s roadmap and Intel’s roadmap, and Steve made the obvious decision. It will set back both IBM and AMD years and years to catch up. All this of course is straight from my source at Intel who is one of the chief engineers involved in the project.
Zupchuck:
I’m one of those ‘advanced amateurs’ when it comes to instruction sets on CPUs, and our conversation is hitting the bleeding edge with what I feel I can say with 100% certainty. With that admission, I do know of one major feature difference between RISC (PPC) and CISC (x86) processors; the way they address the ‘words’ that tell them what to do is completely the reverse of each other (known as the Endian Problem).
Basically, x86 CPUs take a number like “one hundred and sixty nine” and reads it “961”, while PPC reads “169”. The reasons why are very esoteric, which is how they wound up naming it the “endian problem”. The term is based on Gulliver’s Travels; the Lilliputions are at war with some other group because the other group insists on eating their hardboiled eggs from the big (i.e. wrong) end first. That’s about the size of the performance rational between adressing instruction words one way or the other, although PPC words – because they’re in a more humanly recognizable order – were easier for people to code for at first. Nowadays, with automatic compilers for the different software languages, it’s pretty much a wash. However, the legacy difference between the two CPU types is still with us, and software and OS’s have to take it into account before they will work. Beyond all the other reasons why x86 and PPC differeces require Rosetta’s and VirtualPCs, that’s one too.
Also, the link you have here indicates that instructions sets are still, and perhaps always going to be, a consideration. Quote:
“Intel had originally decided to … drop x86 compatibility with the 64-bit generation, by introducing a new architecture called IA-64. IA-64 technology is the basis for its Itanium … processors. IA-64 is not software compatible with x86 software natively; it uses various forms of emulation to run x86 software.”
PPC is just as different from x86 as Itanium is from x86, and translating between them will always require some specialized hardware and/or software, which will in turn exact a performance penality that the average user won’t want to put up with on a day to day basis (unless they have no choice of course). However, the article says there is hope for emmulation/virtualization:
“Intel and AMD have both announced that future x86 processors will have new enhancements to facilitate more efficient virtualization.”
So maybe the future will be a little less hardware dependent, but I wouldn’t count on any ol software or OS being able run on any ol CPU, any time soon.
Apple’s own actions would indicate that they don’t believe they’ve transended CPU dependency. For example, as good as universal binaries – a software innovation – are, they in no way are unshackling OSX from a given CPU architecture; they just make it easy for the developers to ‘switch shackels’. If it truly didn’t matter which CPU Apple was using, I believe we would never had an announcement of a major CPU shift in the first place.
Eric24601:
Sure hope you’re right, but – no offense to your source – I personally need to see more proof in the pudding. So far, both AMD (x86 specific performance) and IBM (cutting edge CPU design) have been making Intel look slow-footed. The roadmaps anyone has actually seen doesn’t favor them in any overwhelming way either. And, if one reads the ;legal section of the newspaper recently, Intel seems more interested in leveraging their market power than in taking the lead in CPU power, which is pretty consistent with their history for the last decade.
I am crossing fingers though.
One word: Williamette.
Intel has a history of first run production crap. The Williamette P4 was out performed by the older Tualatin P!!! core. AMD beat Intel to 1Ghz because Intel’s tape out on first run production sucks. It takes them time to get their fabbing process right. Don’t get your panties in a bunch thinking this news is the next great Apple savior. It wont happen that way. Check the history.
Intel’s original successor for Prescott was called Tejas (which is spanish for Texas). Tejas was cancelled after the Prescott power fiasco and its successor is Conroe. The only reference to Conroe on Google that I could find references the seat of Montgomery County, Texas. Which is where I happen to live.
” width=”19″ height=”19″ alt=”smile” style=”border:0;” />
http://endian.net/details.asp?ItemNo=3613