Although Apple hasn’t announced which Intel chips will power the upcoming Mactel systems, the most likely suspects are microprocessors that have been dubbed Yonha, Celeron D 3515, Conroe, Woodcrest, Smithfield and Merom. For the rest of the week, we’ll offer a look at the chips and what they might mean to Mac users.
The question it did raise is will OSX remain a 32bit OS on just the powerbook/ibook/mac mini range?
Anon
I betcha a lot of ISV would be downright pissed if Apple transitioned from PPC to IA32 and then again to x86_64. Maybe Intel will release a really good 64 bit mobile chip, so that Apple can have a straight 64 bit OS.
But what do I know? đ
I betcha a lot of ISV would be downright pissed if Apple transitioned from PPC to IA32 and then again to x86_64
Not really since the whole advantage of x86_64 is it can run unmodified x86_32 code. The same reason you didn’t see any software developers complain when the G5 was released.
I think it’s pretty much a guarontee that at least the iBook line and probably the PowerBook line as well will be 32-bit initially for the reason stated in the article – battery life. Most laptop users would gladly trade 64-bit capabilities in for an extra hour of batter life.
I think it’s pretty much a guarontee that at least the iBook line and probably the PowerBook line as well will be 32-bit initially for the reason stated in the article – battery life. Most laptop users would gladly trade 64-bit capabilities in for an extra hour of batter life.
That’s strange, though. The (supposed) reason Apple dumped IBM is because they couldn’t deliver 64-bit (G5) processors for notebooks, correct?
Please, please, please tell me that Apple wouldn’t ever consider putting Celeron chips in Macs. That would just hurt.
apple wanted faster laptop chips but they are not getting 64 bit chips from intel from a long time.
That kind of kills the we switched to intel because IBM could not make a mobile G5 story jobs gave. Apple switched because apple wanted to switch.
âWe made a conscious decision not to include itâ because of the impact on battery life, Eden said. Intel said it will release a 64-bit chip for notebooks when the market ârequiresâ it, CNET says, adding that what ârequiresâ means is a source of debate.
Boy intel still doesn’t get it do they?
I’ll definatelly kill myself if there is a Celeron in my Mac
Isn’t Yonah EOL ? If Apple was to use Yonah, would it expect till next year to do it ? I bet they’ll never use Yonah, but some CPU that isn’t in production yet.
I think Intel have got it right – whats the point of going 64 if it reduces battery life and not much speed increase, what is the point?
Wow…
Now, that’s something to dream about đ
> Isn’t Yonah EOL? If Apple was to use Yonah, would it expect till next year to do it ? I bet they’ll never use Yonah, but some CPU that isn’t in production yet.
Yeah IMHO you’ll see the 64 bit dual core Conroe or Woodcrest processor in the desktop Macs, I have NFI what you’ll see in the notebooks, but a dual dual core Intel Mac running at 2.5 GHz per core that’ll be a nice upgrade and you can sign me up once the hardware hits Rev B.
> That’s strange, though. The (supposed) reason Apple dumped
> IBM is because they couldn’t deliver 64-bit (G5) processors
> for notebooks, correct?
Not especially correct. The 64-bit aspect of the G5 was largely unimportant, since OS X’s 64-bit support is really rather limited. Where it is practically exploitable it isn’t particularly meaningful to laptop customers. Apple wanted a processor that they could put into laptops that would be performance and power competitive with the Pentium M. This wasn’t being met to their satisfaction by Freescale with the 750 nor IBM with the 970.
Right in the developer documentation for the Mac/x86 it’s referred to in terms of the 32-bit ISA. There may be a further transition to x86-64, but that isn’t an immediate concern. Since most of the libraries one associates with development on the Mac aren’t 64-bit clean right now, this isn’t overly surprising. A little disappointing from an idealistic perspective, but not surprising, and probably not impractical.
maybe it’s the chips that steve jobs had in his macs when he demonstrated them or the chips that are in the developer pcs? I think it will be industry standard.
Pentium M, Xeon, Pentium 4
– Not 64bit?
– Shared cache with shared bus is Better then seperate cache and seperate buses? ( Doesn’t shared bus == Bottleneck? )
Someone at Intel smoking something?
If this is what Apple is switching to it might as well have stayed with IBM.
Like a previous poster asked, adding more registers and therefore a more efficient architecture is going to increase energy requirements?
– Well the chip will reach higher performance more easily with less register swapping so it probably can do more real sustained work, so on video compression it might use more power. But for general processing( word processing ) it might use less!
I wonder why people keep asking for a 64bit laptop processor from Intel.
I honestly can’t see that there is a business case for 4+ gigabytes of memory in a laptop computer.
So honestly what is the point? why do “geeks” keep asking for it? Besides most of the Cocoa framework/user interface code in Mac OS X is 32 bit and will remain so for a long time.
Any normal usage will not benefit noticably from a 64bit processor. If you do big-number math or use gigantic databases then 64bit CPU’s are your number.
I can’t wait for my improved PB 13″ widescreen (I can dream). The Pentium M chip completely rocks and if we are lucky then Yonah will be even better. If we could get 8+ hours of battery on a PB it will be the best thing since we invented laptop computers
Later guys
I hope the new iBooks do not have shared video memory.
Most laptop users would gladly trade 64-bit capabilities in for an extra hour of batter life.
Hardly. From what I’ve seen, the average notebook user will sacrifice more than half the battery life and their left nut for even a minor performance increase.
For every notebook that jumps through hoops to give you an extra 5 minutes of battery time, there are a dozen that give you yet another P4E with ATI X800 graphics and a 17″ LCD burning through the battery in an hour.
Apple didn’t drop the 1.67 GHz G4 to go with a 1.4 GHz Celeron. You can bet the xBook will be at least a 2.5 GHz PentiumM with go6600 graphics. If Intel has the 64bit version of the PM by then, you can bet that’ll be what Apple uses.
with a full 64 bit os and full 64 bit apps you generally see about a 30% increase in performance over the identical specs running 32 bit (depending somewhat on the application)
so yes… it has a lot to do with performance, not just with how much memory the system can utulize
I know this is comparing apples to oranges but….
I have experienced this first hand with my Linux installation on my amd64 box, running the same os with the same apps, same gcc, same compiling optimizations, on the same computer, one 32 bit, and one full 64 bit (yes, not one 32 bit app)….. i have seen a substantial increase in general system speed and a phenominal speed increase in compiling times possably cutting them in half.
just some food for thought
anyway…. i also realize that even with the ability to run full 64 bit and with the performance increase, it will take some time to get all the software vendors to jump onboard.
…will have EM64T, according to the Inquirer:
http://theinquirer.net/?article=19149
While the Inquirer isn’t exactly the most reputable source, it’d be foolish to think that Intel would include such massive architectural changes as shared cache in a dual core CPU and not bother to include EM64T.
So all this bitching about IA32 is moot. Intel plans to phase out all their non-EM64T chips by 2007 anyway. From that I think it’s safe to say that we won’t see an Intel-powered Mac shipping that won’t support EM64T.
I am glad you pointed this out. You are correct in the AMD64/X86-64/EM64T case due to it being very register starved compared to only architectures. However for G5 PowerPC that was not the case, so there is little reason for it to go 64bit except for the memory increase. Other people still seem to forget the biggest user visible improvement with the AMD64 arch is the increased number of registers (~ 30% speed improvement), not the ability to access over 4GB ram. Though in the future being able to access over 4GB of ram will be handy as well.
Hardly. From what I’ve seen, the average notebook user will sacrifice more than half the battery life and their left nut for even a minor performance increase.
“From what you’ve seen” indeed. Looks like you see a lot of desknote users – those cheap notebooks are overpowered because they use desktop parts and were never designed to be used as a true notebook. There are indeed lots of ’em, but until now, there haven’t been very many in Macland, and I suspect that this will not change. Expect true mobile specs – power-saving processors, realistic battery life – from the Intel Mac iBooks and PowerBooks.
I can’t help thinking that sticking with IBM and the PowerPC line leading into the realization of the new Cell processors would have been the opportunity and the requirement to actually “Think Different”.
no more 8GB of ram
back to 2GB of ram
I can’t help thinking that sticking with IBM and the PowerPC line leading into the realization of the new Cell processors would have been the opportunity and the requirement to actually “Think Different”.
Seriously, all else being equal, do you think we’d be having this conversation right now if IBM had provided 3Ghz G5’s and mobile G5’s on a reasonable schedule? Blame whoever you want – technical glitches at IBM, Apple’s measly marketshare making rushing not worthwhile, whatever. But once you acknowledge that Apple was stuck spitting out machines with chips *two years* behind their roadmap, what more is there to discuss other than “who’s the next best vendor, what do they sell, and sign me up”?
Well fucking said. There’s nothing more to discuss on Apple’s PPC to x86 switch. Mac fanboys should accept this as a fact and move the hell on. x86 is here to stay, whether you like it or not. And believe me when I say thank God for x86, the over-hyped Cell processor doesn’t even come close in sustained performance like x86 processors do, so Apple wouldn’t even stand a chance by going that down route.
Would someone explain to me how making Yonah 64-bit leads to higher power consumption??
The Roadmap for the PowerPC has been out for years. Like you I don’t understand what the delivery hassles are. First Motorola, now IBM.
But it is also reasonable to assume that the Cell processor has been known about for some time…. the boost in capability also. It seems to truly be an oppotunity that will almost forge its own path. I wonder why they haven’t been designing and developing on it for 3 or 4 years. Apple has been gaining credibilty and market share almost everywhere computers are deployed, and this kind of power boost would have given them a performance leap-frog over the whole field.
Anyway, I continue to watch and listen, maybe I’ll think different(ly).
I had a celeron processor in my first computer (that is, the first one that was actually mine and not a family computer), and the speed wasn’t much different from equivalent computers running on the Pentium 3 chips of the time. I think a lot of people are going by hearsay rather than actual facts, sure there is probably a difference, that is why Celerons cost less; but, the fact is that a Celeron still runs quite nicely. The obsession with having a Pentium 4 (or whatever is comming next) goes back to the age old tradition of bragging rights and felling good about one’s toys.
Really the only people who I can see justifying this dislike for Celeron processors are hardcore gamers, software developers, and graphics artists, in which case the slight difference in speed would be more important.
If you all hate Celeron processors that much then there’s more for me đ .
Some people see the word “celeron” and immediately make a stink about performance. Yes, Celerons are meant to be ‘crippled’ pentiums, but usually, that means taking out speedstep and half the L2 cache. Its not that big of a deal, you can continue typing your PC bashing from you mac without a problem.
Please, please, please tell me that Apple wouldn’t ever consider putting Celeron chips in Macs. That would just hurt. — Roguelazer
/me agrees….
I’ll definatelly kill myself if there is a Celeron in my Mac –Anonymous (IP: —.cable.mindspring.com)
Don’t know as if I’d go that far, but I’ll tell ya this… I’ll NEVER own a Mac with a Celeron CPU. And as it is, I will most likely not buy a newer MacTel. Seriously x86 (even x86-64) is a serious step backwards vs. PPC.
Some people see the word “celeron” and immediately make a stink about performance. Yes, Celerons are meant to be ‘crippled’ pentiums, but usually, that means taking out speedstep and half the L2 cache. Its not that big of a deal, you can continue typing your PC bashing from you mac without a problem. –a pc guy
Every Celeron system I’ve used, has always been shit for performance. I would never own one, Celeron is a waste of money, and IMHO is not even worthy of running a pocket calulator ;-p
–Jed
Glad to see people, and Intel, coming around to the fact that the Pentium M is just about the best x86 anything to come out of Intel since the original Pentium Pro. Die, Netburst, Die.
I SINCERELY want to believe that Yonah will support EMT64, as others have said, and that the first MacTel Powerbooks will use it. Good thing Intel chose to crib AMD; of course, Microsoft didn’t give them a choice. A clean start of MacOS X exclusively on x86-64 would further polish Apple’s image as an innovative leader, even on a commodity platform. Yes, I know that Linux, the BSD’s, and Windows were there first, it’s about the image. Just like USB was around and working before the iMac, but nobody cared until afterward. When the G5 came out, you’d have to BEG Apple to stop trumpeting its 64-bit capabilities. You couldn’t spit and not hit a smug Mac fan, sitting proudly with their 64-bit G5, even though Jaguar couldn’t DO anything with it. Now they have to go back on all that. hey’ve already got the OS working, and tey have time until 10.5 (Leopard). Come on, Apple, lead.
I think most laptop users would trade up to 20% of battery life for 30% improved performance. Honesly, while IANA chip designer, I doubt it’s even that much battery life lost. The only crowd who would complain would be the ultraportable users who are targeted by the iBook and 12″ Powerbook. I think they’ll be able to deal with the loss in exchange for Apple having respectable laptop performance across the board. Besides, there are other ways to reap power savings.
Mind you, battery technology is improving and the graphics and peripheral chip makers (including Intel themselves) are getting smarter about power. Intel won’t be able to claim sole credit for 8+ hours on a single charge when they get there.
–rood
in addition to more registers, it jettison many legacy features such as segmented registers and x87 FPU. so unlike powerpc making it amd64 native should boost performance
“I am glad you pointed this out. You are correct in the AMD64/X86-64/EM64T case due to it being very register starved compared to only architectures. However for G5 PowerPC that was not the case, so there is little reason for it to go 64bit except for the memory increase. Other people still seem to forget the biggest user visible improvement with the AMD64 arch is the increased number of registers (~ 30% speed improvement), not the ability to access over 4GB ram. Though in the future being able to access over 4GB of ram will be handy as well.”
I think people are underestimating the utility of a 64-bit machine. The RAM limit is a very real one. The 4GB virtual memory limit means, that for all intents and purposes, the real usable RAM capacity is 2GB. If your OS uses a 3GB/1GB split, that leaves you with some room to map graphics RAM (which is by itself up to 512MB these days) and device registers. Of course, even at 2GB, a 32-bit OS has to use tricks (like highmem on Linux) to use anything past the first 1GB or so from kernel space.
These days, 1GB is quickly becoming the standard memory configuration. Power users often have more. 2GB of high-quality DDR400 can be had for only $250. It’s not at all unreasonable to believe that it’ll only be a couple of years before a significant number of users are itching to put more than 2GB of memory in their computer. Especially given their graphics pro base, it’d be silly for Apple to go through the agony of the x86 transition only to force a 64-bit transition a few years down the line.
Yonah will have three important new features. The first is Intel Digital Media Boost, an instruction set for rich digital multimedia creation.
Oh please, not another instruction set extension. Great how the article doesn’t provide any actual information on it. So what kind of instructions does this actually mean? I thought Intel x86 already has its fair share of media extensions with MMX and SSE/2/3.
in addition to more registers, it jettison many legacy features such as segmented registers and x87 FPU.
While that is nice, it doesn’t actually gain you anything. That’s because the hardware is still on the chip anyway to support compatibility mode. Of course they might one day throw out compatibility mode, but considering even braindead 8086 segmenting is still in there, chances of that happening anytime soon are zero.
Besides, the FPU is still available in 64-bit mode, it’s just that Win64 (and Linux?) don’t save its registers across task switches anymore. And segment registers are still there as well and can be used through prefixes, but instructions don’t use them implicitly anymore.
blah blah Cell blah blah
You might want to read up on the Cell’s disadvantages as a desktop processor.
As long as Apple maintains the same level of quality in their hardware and software implementation then you could run it on a celeron 667 for all I care. Performance is a only a part of overall usability.
I doubt however that they (Apple) will use the celerons, not for technical reasons, but simply from a marketing standpoint.
Shared cache with shared bus is Better then seperate cache and seperate buses?
Yes, shared cache of 2 MB is better than two separate caches of 1 MB, because it allows the whole 2MB to be utilised when only one thread is running or when one thread doesn’t use a whole MB.
Doesn’t shared bus == Bottleneck?
Not if it has twice the bandwidth (the article doesn’t say). Or if the separate buses are under-used anyway, which is often the case because the (still separate) level 1 caches fulfill most requests.
In any case, the main problem with memory is latency, not bandwidth. You can always make your buses wider, but you can’t make those pesky DRAM cells faster.
Yonah will carry 151.6 million transistors and integrate a 667 MHz front side bus
Not exactly a desktop replacement what seems to be the trend nowadays.A mediocre desktop these days has a 800MHz FSB.
The 2MB L2 cache isn’t the only tweak apostel.There’re a lot AMD X2 CPU’s with less than the 2MB L2 cache but still impress with smokin overall performance (per watt consumed).
People that bash Celerons without reason clearly have never read current reviews, nor can they read specs. A Celeron M 370 chip (which a derivative of would go in an iBook, if they were to use Celerons at all) is 1.5Ghz, w/ 1MB of L2 cache. It’s last year’s Pentium M.
I don’t see why that’s so horrible, and I would totally buy an iBook for $400 less than a PowerBook if this was one of the only differences (just as now there’s very little differences between the iBook and 12″ PB)
cleansheet redesign is what i was suggesting.
from wiki
“As of January 2005 the product does not yet exist, but the assumption has been made by several AMD-watchers, including a Japanese site named PC-watch 1, that it will come out around the end of 2005 or the beginning of 2006. Development of this new generation of CPUs started about March 2003.2 In contrast to past generations of CPUs by AMD and Intel, this new generation of CPU will probably take less time to develop, due to optimizations in the development process by both Intel and AMD. Some sources indicate that development will last about two years; if that is true, the K9 can be expected around Q1 2006.
When considering possible designs, it is perhaps worth noting that AMD64 when combined with Windows XP Professional x64 Edition, allows for the elimination of the physical FPU unit, MMX, and 3DNOW! instructions. Therefore once 64 bit Windows is a mainstream platform, the potential exists for a clean sheet redesign of x86 hardware. With this in mind, the K9 could prove to be a surprising chip, looking much more like a classic RISC processor, than has traditionally been the case with x86 systems.”
I think people are underestimating the utility of a 64-bit machine. … it’d be silly for Apple to go through the agony of the x86 transition only to force a 64-bit transition a few years down the line.
I think you’re overestimating the difficulty of the “transition.” Assuming Apple handles EM64T/AMD64 much as they handled the G5, what exactly do you think will be involved, other than a new point release of the OS and new apps to take advantage of it?
A clean sheet K9 is a nice thought but I just don’t see it happening. Compatibility with 32-bit software is simply too important. And as long as Intel’s EFI initiative hasn’t succeeded in replacing bad old BIOS even the truly ugly 16-bit stuff has to stay in.
My Desktop does not have to be a Cell. My Desktop could still manage my workflow, and direct the processing requirements of the job at hand to connect with the capabilities of the Cell…… go MicroKernals!
I don’t really understand what you’re talking about. So you’d want a separate box with a Cell processor in it?
The only way the Cell would possibly have made sense for Apple would have been with a G5 as its “PowerPC Processing Element”. But I guess even then they’d rather have a second G5 core instead of those difficult to program SPEs.
And none of this would have solved Apple’s problems with laptop processors.