“A deluge of reader mail has opened my eyes to some very sound reasons why the Motorola PowerPC chip’s days may be numbered I love nothing better than stirring up a hornet’s nest. And that’s just what I did with my recent column about whether Apple should abandon its current PowerPC microprocessor for a Pentium-family chip, like those that power most Windows PCs (see BW Online, 9/11/02, “Mac and PC: Ne’er the Twain Will Meet“).” Read the new editorial at BusinessWeek. Update: iSync beta!On a related Apple story, OSAlert reader Prognathus sents us a discussion link with a detailed description of Jaguar as a counter-productive version of MacOS, written by a MacOS 9.x advocate.
I still don’t see why people still dream about this. Yes I would love to be able to run OS X on my Pimped out PC, (as well as my mac’s). But I stand firm is saying that Apple will never go X86, they’ll choose any other alternative before X86.
Every time one of these articles comes up, I try to practice will power not to come out and troll. But this particular topic has been hit several times, this is an unrealistic dream people, while Apple may have considered this alternative in the past, didn’t we just hear another article about IBM and Mac teaming together?
I think the last line says it all.
Apple’s marketing is still so anti-intel/x86 that they would have to basically go against years of their “pentium crushing” marketing with PPC. It’d be too embarassing really.
The one argument people keep fielding why Apple would never chose to go Intel is that they would lose hardware sales.
It has been reiterated quite some times that nothing would keep Apple from providing “special” hardware with their OS, disabling to run their OS on “stock” hardware. If you want to run MacOS, you’d still have to buy the HW from Apple.
This is possible (in fact, Eyetech is doing something similar to the AmigaOne), and I have yet to hear why this wouldn’t be the best of two worlds for Apple: Faster CPUs for the MacOS, HW sales for the revenue.
i have never thought of that aspect, but i agree, Jobs would have to eat a lot of crow if he went x86.
I’ve said it before I’ll say it again. If Apple were going to make a processor switch the easiest route is to some sort of modified Power4 chip. There are consistent rummors that Apple is looking to make this switch and IMHO it is the most likely.
If that fails the Itanium 2 makes a great deal more sense than the x86. And before someone comes in with the price argument; if you look at Intel’s price list Itanium 2’s are no more expensive than Xeon Pentiums with similar cache. That is bring the cache down to Pentium 4 levels and the processor is no more expensive than the Pentium 4. Its the cache not the chip that is making these chips seems expensive and Intel can fix that in weeks / months.
http://www.intel.com/intel/finance/pricelist/
Further were Apple to offer the switch to Itanium 2 I could very easily see Intel being willing to move the chips at cost or even a loss to put pressure on Microsoft.
1) The Pentium 4 x86 is a bad chip that runs very quickly; not a good chip. The biggest advantage of the x86 line is support for legacy code that doesn’t run on Mac anyway. Its not just that Steve Jobs would need to eat crow, he would need to back off from completely true claims about why x86 stinks.
2) There is no substantial price advantage to the x86 for the kinds of customers that Apple wants (i.e. not the pricewatch crowd).
3) The chip doesn’t offer Apple any advantages:
a) x86 with its terrible power consumption would hurt Apple laptops quite a bit in terms of battery life and the laptops are strong margin products.
b) On the powermacs going dual seems to be a very workable strategy. Further dual systems actually work out to be more pleasent environments (i.e. they feel faster) then single processor environments under high CPU loads. Even Motorolla can probably handle getting the bus speed up on the next generation of G4s.
c) The Apple XServe can afford to underperform on CPU because its real aim is ease of administration
d) The eMac/iMac crowd is not unsatified with the performance
If Apple is going to ask their developers to port; its going to be to a much superior platform. Mac users tend to be not terrible price/performance sensitive and very conservative. They aren’t going to be happy about a switch unless they are getting a great deal more value or performance.
It is a very will know fact that Apple is working with IBM on future processors. People would not be interested in a proprietary Intel box and it’s just suicidal for Apple to have OS X run on the current standard Intel Boxes.
Hate to say it but I forsee problems for Apple. People are just not buying desktops (Mac or PC). OS X is not helping Apple any up to this point.
Oh well, time will tell.
ciao
yc
…but I agree with jbolden1517 that the “pure” x86 isn’t the best of the worlds, unless they have no other way out.
IBM looks as the first choice, IBM Microelectronics is trying to prove they can have more customer than IBM Unix group. And – yes – Apple can do their “Pentium sux” mantra again in style.
AMD Opteron looks like the second choice. AMD needs to have a major coup against Intel, to show the market they can be a reliable chipmaker and not only the choice of Taiwanese mobomakers; also needs a showcase to the Opteron power. And AMD isn’t Intel – has anyone ever seen Apple saying “AMD sux”?
Itanic 2? IMO no way. Jobs isn’t going to eat his own words about Intel. It’s too much for his Galaxy-sized ego.
> I have yet to hear why this wouldn’t be the best of two
> worlds for Apple
I know some good reasons:
1) Many currently available PPC Macintosh software must then be ported over to the new x86 platform.
2) Offering an alternative solution to x86. Believe me or not, there are many people who prefer PPC CPUs for *many* different reasons, either for their size, coolness (literally speaking), PPC assembly, etc.
3) As being an important partner together with IBM and Motorola, a CPU switch would for Apple be like shooting into Apple’s own foot.
IMO x86 isn’t a good long term option. So what is a good *long* term option then (As PPC isn’t one a good one either)?
Enter Tao’s Virtual Processor! http://tao-group.com (Native execution performance/small load-time translation penalty)
IMO the proper road to follow is to have easily portable platform specific OSes. Like i.e. x86 AmigaOS, PPC AmigaOS or MIPS AmigaOS and *then* offering a binary identical software layer on top.
Now when a revolutionary new CPU comes along OS developers would simply need to port AmigaOS to this platform including the intent layer and voila all previously available software will become available for consumers and developers to use.
I don’t think Apple would ever go pc compatible. The only option I see out of this is a custom intel/amd procesor (most likely 64 bit) most and a proprietary motherboard created by apple. Whichever direction they go, it probably wouldn’t really be a step towards bringing Osx to the pc someday.
Not this tired subject again. Sheesh!
dear god a apple x86 thread without a single “Apple would make more money switching to intel” arguement, praise jesus.
anyway this article is saying stay with ppc for all the wrong reasons, and looking at roadmaps, and other published information, PPC is not going to die anytime soon.
IBM doesnt invest tons of money into expensive cpu fabs to waste money. Apple is a major PPC buyer, they can also ship PPC linux systems, or AIX. Im extremely happy with this x86 competition coming.
Quote: “The big potential losers if Apple should switch chips would be software developers. They would be forced — perhaps for the second time in two years — to rewrite their programs, this time to make them work with a Pentium-based Mac.”
Well, porting software from PowerPC to the Intel Architecture should involve not much more than recompiling! So this statement is false, I believe. Software developers do not have to rewrite their programs unless they were written in pure assembly. Maybe some minor tweaks, but not more than that.
Someone needs to do the math a bit better.
If today the possible next gen type of processor in limited production is already being sold in a box from IBM for $12,000 then how much can a mass produced item be sold for taking into the fact of return on RD on a sum of the years? About the starting price for a newly introduced Mac!
Why doesn’t anyone believe that the x86 is a (blank) toy? There are so many design problems that have been patched and patched over and over again that it is just plain stupid to waste time on such junk!
Don’t tell me it went to the moon again please. That was a limited and HIGHLY tested single use that didn’t use DOS.
These chips are both bery expensive.
Motorola isn’t doing much for Apple in terms of chips, they are still producing embedded processors. What makes you think Intel would produce a cheap low end Itanium 2 for Apple? They really don’t have that much reason to. Its not that easy to just remove some of the cache without severly crippling the chip. The same is mostly true of Power4.
Now according to the IBM-Apple deal, IBM may be producing a workstation version of the Power4( it may be similar, but it won’t be the same as the server version), but it will be for their own workstations, and Apple, probably not just for only Apple.
Apple has so little market share that making chip for just Apple it suicidal.
If Apple wants to change architectures, they’ll probably have to settle for a processor that someone else is already making. If they do choose a really expensive and powerfull processor, what do you think prices of Macs are gonna do?
Come on guys, that ‘Apple switch to x86’ discussions are getting boring. They are without any sence, reality and are not helpful nor they are OSAlert. Maybe Euqenia sould switch the name of this site to AppleSwitchToX86Cant. Even more boring is the flamewar that starts alle the time about this discussion. Do you think you will change anything? Do you think S.Jobs reads OSAlert an thought – hey intelligent people here – nice suggestions! LOL!
And for the article about OS X:
Wake up boys and girls – OS9 is dead. All apps are on OS X now or are on the way. And this is a good thing. OS9 is old, ugly and has a bad OS design. It is even not a real OS – better call it programstarter. Apple does a great job with OS X. It is fast, reliable and beautiful. It helps apple to get back the lost marketshare. In the MS/Intel world no one will flavor Win3.1 over WinXP. That is the equal relationship between OS9 and OS X.
Ralf.
Motorola is an embedded company. They’re good at making chips for small, not-very-powerful devices. Think Cisco routers, handhelds, and theoretically even laptops; devices where low power consumption and high battery life, not megahertz and frame rates, are key. This might change eventually, but for this reason Apple will probably continue to use Motorola procs on the low end for a while.
The high end, however, is a different story. Motorola is completely gone from the high-end. IBM makes really fast processors, but they’re almost too fast; way beyond what the average desktop user will ever need – these are more like mainframe-level processors. AMD and Intel make processors that would be the right price/performance mix, except they run on the wrong platform.
But… AMD and Intel use RISC internally. Both companies used processors at some point that broke down CISC instructions into RISC microcode, a la Transmeta in hardware, that the processor actually ran. They might even still be using this technology. Either way, it would be painfully easy for AMD to take a Hammer or Intel a P4/Yamhill, strip the CISC overlay, switch to a 4- or 7-stage pipeline from the current 14/20, and ship Apple a high-power processor. Not only would this processor be free of x86 legacy while still being x86 powerful, so that it wouldn’t be hard at all for Apple to port Jaguar to this processor; but because of its x86 roots, Apple could take the stripped-off CISC emulation layer, turn it into software, and instantly have a superfast low-level x86 emulator. Combine that with WINE, and you have a computer that can run both Mac and PC applications faster than a Windows box!
Unfortunately, I doubt Apple has explored the possibility. But an AMD-powered RISC box is the best of both worlds, as Apple would get the most powerful processor on the market, and the capability to run apps designed for the other 95%, while continuing to use the clean legacy-free architecture that makes them popular among power users. Were Apple to switch to Intel, they might as well declare Chapter 11, because they’ll simply turn into a higher-priced, underpowered Sony; were Apple to remain with Motorola, attrition would soon stop them in their tracks.
This would be a really funny situation. Itanium has a couple of major problems:
1) Its got large instructions. EPIC instructions are larger than RISC instructions, which are (overall) larger than x86 instructions. This makes the 1.5MB of cache on the Itanium something that’s actually necessary, rather than just gratuitous. Getting rid of a lot of that cache to save money would lead to a significant performance hit. Though, getting rid of the cache *would* work. Without the cache, the normally 221 million transistor Itanium2 is only about 60 million transistors.
2) It’s hard to compile for. While the benchmarks show excellent performance for Itanium2, thats for scientific processing and with HP’s compiler. Desktop program code is much higher level and much harder to parallelize and compile-time. Also, HP’s compiler is a far cry away from GCC, which Apple would have to use for OS X.
3) Its hard to design for. Motherboards using Itanium have to be a lot more complex, which jacks up the price. Its got a 128-bit bus (most PC boards have a 64-bit bus) which increases the number of traces the manufacturer has to run through the motherboard. While its not impossible (the nForce, after all, also has a 128-bit bus) it is harder will make motherboards more expensive.
4) It’s hard to scale. Architecture is nice, but in the end, its pure speed that matters. In scientific processing, its relatively easy to keep a dozen execution units busy, but that’s impossible to do in desktop-type code (except games!) Thus, narrow but fast architectures are inherently better than wide but slow architectures in that area. If you take a look at the SPEC marks, you’ll see an Itanium 1 GHz is 50% faster than a 2.5 GHz P4. Wow, that’s quite impressive. *BUT* Intel has already demoed 4.7 GHz P4s. The competition with AMD is racheting up P4 speeds like anything. How long do you think that 50% lead is going to hold? Intel won’t let the P4 overtake Itanium2 for server/scientific apps, where its bus architecture and memory bandwidth will come into play, but for desktop apps? The P4 is already faster than almost all RISC chips once you take them out of their fancy huge busses and multi-core implementations.
Shouldn’t apple be growing their business? Why must they be depending on hardware in a world where hardware prices are ever getting cheaper? They’ve had more than enough time to diversify and make the switch!
Apple should switch to Intel. People should be able to grab a copy of OS X and install it on any whitebox. Who wouldn’t like to run OS X? Grow the userbase, and good things will happen.
I agree it will be a bold and risky move, but unless they do it, they too will go the way of Sun. Their market share has already dwindled from 5% to 2.something.
> Why doesn’t anyone believe that the x86 is a (blank) toy? There are
> so many design problems that have been patched and patched over and
> over again that it is just plain stupid to waste time on such junk!
The x86 box is just a toy that:
1. Outperforms Apple dollar for dollar.
2. Outsells Apple immensely.
3. Can actually run software found in most electronics stores.
4. Can actually use hardware found in most electronics stores.
5. Can be bought very cheaply if necessary.
6. Gives me many more choices.
Man, I am glad that I am getting rid of my x86 junk in favor of a nice, shiny, new, efficient, well-designed, cool, PPC box that has little practical use for me. But the design will make me feel really good inside, and I will like the way the case looks as I just sit there thinking of all the things I could be doing on my PC.
Oh, and for the intellectually challenged, that was sarcasm.
Oh god. I hate people complaining about x86 being crufty. There is almost no cruft left in current x86 processors. It’s all been “depricated.” Its still there for you to use, but god help you if you try, because those weird bits haven’t been optimized in years. Some specifics:
1) Segmentation. Segmentation is no longer with us. The only use of segmentation that most modern 32-bit OSs do is set up a code and data segment to map all of the address space, and segmentation is never dealth with again. Once those two segment registers are loaded, it doesn’t slow down the processor either, as AMD/Intel has probably optimized out the common case of software that doesn’t touch the segment registers. Even system calls don’t really use segements anymore, because the sysenter/syscall instructions hardwire a flat address space.
2) Variable-length instructions. From a programmer’s point of view, variable length instructions are rather nice. Unlike RISC machines, ASM programmers (and linker writers) don’t have to deal with complicated tricks to full addresses into instructions that don’t fit in a fixed-length instruction. And the processor doesn’t really care, because by now the x86 decode units on the Athlon and P4 are so beefy, they don’t impact performance much. Throw in the trace cache, and variable length instructions turn out to be something of a win, because more code can fit an a given size cache.
3) Weird instructions. The instructions in x86 that deal with binary-coded decimal have not been optimized in a decade. They sit in a microcode library somewhere and don’t affect anything unless someone is stupid enough to use it.
4) Lack of registers. This one does hurt a bit. But register renaming helps this out a lot.
5) Floating point stack. Only impacts code size in Athlon processors (instructions to move operands around on stack take 0 clock cycles, just code space) and on P4s, GCC can be configured to use SSE for floating point instead. Problem solved.
And all of this really doesn’t matter unless you’re coding in ASM. Even if you’re coding in low-level C, x86 is a lot nicer than most RISC architectures because its “weirdness” allows more programmer friendly constructs. What matters, in the end, is that x86 processors are insanely fast and ridiculously cheap. And let’s nip this power issue right now. 50W is nothing. Unless you have a flat-panel and no lights in your work-room, the processor’s power usage is the least of your concerns. Cooling, also, is not a concern if you do it well. Current Dell P4’s have a giant low RPM fan blowing across a huge (fanless) heatsink. They also have a high-end (read: quiet) power supply. In that machine, the harddrive is a whole lot louder than anything else.
And let’s nip this power issue right now. 50W is nothing. Unless you have a flat-panel and no lights in your work-room, the processor’s power usage is the least of your concerns. Cooling, also, is not a concern if you do it well. Current Dell P4’s have a giant low RPM fan blowing across a huge (fanless) heatsink. They also have a high-end (read: quiet) power supply. In that machine, the harddrive is a whole lot louder than anything else.
In general with battery Mac laptops give about an extra 2 hours of battery life over PC laptops. That’s a huge difference.
This would be a really funny situation. Itanium has a couple of major problems:
1) Its got large instructions. EPIC instructions are larger than RISC instructions, which are (overall) larger than x86 instructions. This makes the 1.5MB of cache on the Itanium something that’s actually necessary, rather than just gratuitous. Getting rid of a lot of that cache to save money would lead to a significant performance hit. Though, getting rid of the cache *would* work. Without the cache, the normally 221 million transistor Itanium2 is only about 60 million transistors.
Agree with all. My point is that the Itanium 2 is actually not an expensive chip it runs the same price (san cache) as the P4 even though the volume discounts are lower and the markup much much higher. The low end Itanium 2 is going for a bit over a grand. Mass produced even with 1.5 megs of cache that’s probably going to come in like $700; let Intel cut prices to put pressure on Microsoft and now we are down to $500 and that’s not a serious problem. Cut the cache and you can cut the price to $200 easy. Cache is always a good thing and IMHO 1 meg of cache Itanium 2 for $500 is doable for Apple for the Powermac / Powerbook line.
2) It’s hard to compile for. While the benchmarks show excellent performance for Itanium2, thats for scientific processing and with HP’s compiler. Desktop program code is much higher level and much harder to parallelize and compile-time. Also, HP’s compiler is a far cry away from GCC, which Apple would have to use for OS X.
GCC has been out for years for the Itanium line. Its not going to cost Apple much to improve GCC on Itanium. Heck Intel might absorb the cost as part of the deal (and they have lots of experience in writing pretty good compilers). I’d consider this a non issue.
3) Its hard to design for. Motherboards using Itanium have to be a lot more complex, which jacks up the price. Its got a 128-bit bus (most PC boards have a 64-bit bus) which increases the number of traces the manufacturer has to run through the motherboard. While its not impossible (the nForce, after all, also has a 128-bit bus) it is harder will make motherboards more expensive.
Agree here. Lets call it an extra $100-200 at worst.
4) It’s hard to scale. Architecture is nice, but in the end, its pure speed that matters. In scientific processing, its relatively easy to keep a dozen execution units busy, but that’s impossible to do in desktop-type code (except games!) Thus, narrow but fast architectures are inherently better than wide but slow architectures in that area. If you take a look at the SPEC marks, you’ll see an Itanium 1 GHz is 50% faster than a 2.5 GHz P4. Wow, that’s quite impressive. *BUT* Intel has already demoed 4.7 GHz P4s. The competition with AMD is racheting up P4 speeds like anything. How long do you think that 50% lead is going to hold? Intel won’t let the P4 overtake Itanium2 for server/scientific apps, where its bus architecture and memory bandwidth will come into play, but for desktop apps? The P4 is already faster than almost all RISC chips once you take them out of their fancy huge busses and multi-core implementations.
This one I have to disagree. The x86 line is very complex. Chip speeds are hitting all sorts of quantum limits. I think getting to a 3ghz Itanium 2 is going to be much easier then getting a 10ghz Pentium 4.
>> “Itanic 2? IMO no way. Jobs isn’t going to eat his own words about Intel. It’s too much for his Galaxy-sized ego”
If he took $150M USD from Bill Gates, then he’ll do whatever it takes to keep Apple moving forward.
>> “IMO the proper road to follow is to have easily portable platform specific OSes. Like i.e. x86 AmigaOS, PPC AmigaOS or MIPS AmigaOS and *then* offering a binary identical software layer on top.”
Or they could use a Transmeta chip to emulate the PPC architecture, but neither route would give sufficient performance.
>> “it would be painfully easy for AMD to take a Hammer or Intel a P4/Yamhill, strip the CISC overlay, switch to a 4- or 7-stage pipeline from the current 14/20, and ship Apple a high-power processor. Not only would this processor be free of x86 legacy while still being x86 powerful, so that it wouldn’t be hard at all for Apple to port Jaguar to this processor; but because of its x86 roots, Apple could take the stripped-off CISC emulation layer, turn it into software, and instantly have a superfast low-level x86 emulaton”
How fast is “superfast” software emulation?
I hate it when people start attacking processors instead of discussing them.
Nobody knows what Apple’s going to do. Things like the variation of the Power4 sound good, but who knows? Apple’s trying to get people to move over to OS X. That’s number one right now. Perhaps when that is pretty much accomplished, Apple may look at options. Apple can string things out awhile longer. They can do some things in this area. Getting a much faster bus would help alot, then DDR RAM would be of real help then. Enhancing the effectiveness of dual or even quad processors would help. Continuing to get OS X optimized and speeded up would help. If they can get Motorola to keep bringing out incremental speed increases and do these things, then Apple can hang on until OS X is Apple’s only OS.
All this is well and good, we’ve been sooooo down this road.
http://www.osnews.com/moderation.php?news_id=1828
Humor is not allowed.
Moderators and columnist can troll us with all this … yes, yes, I know “If you don’t like it …
But I DO like this site, alot – artist. So sensitive.
Shall we move on?
With a wee bit of humour allowed?
But despite all its short comings it is an architecture that is being developed at an amazing pace, for market apple is in there is nothing to match the current X86 offerings. G4 at 1.25 Ghz may be very fast, but not as fast as the best X86 and the gap is getting bigger not smaller.
Who cares about the design if it outperforms!
If the betrayal on the Intel platform is bad, think about the culture death that would occur if Apple went to Intel and DRM.
In one move, Apple could make their entire culture, their ethos, everything represented by the ‘1984’ ad, make it all into one giant instance of corporate doublespeak.
At least there is a new slogan ready to go:
Everyday, Apple computers make people easier to use.
(or)
We give every citizen the power to be part of their government’s digital hub.
#p
In general with battery Mac laptops give about an extra 2 hours of battery life over PC laptops. That’s a huge difference.
How much battery life do iBooks and TiBooks get? My Compaq 2800T, which is a 1.8GHz P4 with a 15″ UXGA screen and 64MB Radeon 7500 gets 3.5 hours even when on wireless.
let Intel cut prices
to put pressure on Microsoft and now we are down to $500 and that’s not a serious problem. Cut the cache and you can cut the price to $200 easy. Cache is always a good thing and IMHO 1 meg of cache Itanium 2 for $500 is doable for Apple for the Powermac / Powerbook line.
Utter nonsense.
Itanium2 for notebooks?
Cut the price down to $200?
Clueless!
Intel will NOT produce a special Itanium version for Apple. They don’t care a fig about 2% of the market.
They will not produce a version which may be used in markets in which the P4 is prominent.
Even if modified Itaniums were only used in Apples, they would be benchmarked against P4s with apps like Photoshop etc.
This is something which Intel would not allow to happen.
As for having Itaniums in Powermacs, I’d like to see the size of the battery!
GCC has been out for years for the Itanium line. Its not going to cost Apple much to improve GCC on Itanium. Heck Intel might absorb the cost as part of the deal (and they have lots of experience in writing pretty good compilers). I’d consider this a non issue.
The best compiler-writers at HP and Intel have been struggling for the best part fo a decade to improve the performance of the Itaniums.
The performance of GCC at compiling Itanium code is pitiful, compared to the Intel and HP compilers.
Contrary to your assertion, Apple have almost no experience writing good compilers. They can write compilers, but not good ones.
They won’t get very far with Itanium code – it is vastly more harder than optimizing for the PPC.
3) Its hard to design for. Motherboards using Itanium have to be a lot more complex, which jacks up the price. Its got a 128-bit bus (most PC boards have a 64-bit bus) which increases the number of
traces the manufacturer has to run through the motherboard. While its not impossible (the nForce, after all, also has a 128-bit bus) it is harder will make motherboards more expensive.
Agree here. Lets call it an extra $100-200 at worst.
You hafta be kidding. Seriously.
Such a board will cost in the vicinity of $1000.
Not the least reason for the price is it would be produced in incredibly low quantities.
Even if they slimmed the board of features and managed to produce it in less than 8 layers, and got the price down to $500, that is still too expensive for Apple.
This one I have to disagree. The x86 line is very complex. Chip speeds are hitting all sorts of quantum limits. I think getting
to a 3ghz Itanium 2 is going to be much easier then getting a 10ghz Pentium 4.
What?? Are you serious?
I’d say it will be ten times easier to get the P4 to 10GHz than the Itanium to 3GHz.
The reason being that P4 was conceived and designed right from the very start to scale to incredibly high clock speeds.
Whatever is this guy smoking?
It does illustrate the dangers of allowing people who are totally and utterly clueless to post on matters about which they know nothing.
It could, very possibly, be time for Apple to change processors again. But if they do, they need to look forward, not back. Moving to x86 is looking backwords. Not in terms of current performance, but in terms of computer science, x86 is a dead architecture. I admit that it has proved to be durable & extensible – so did the Roman Empire. But neither existed because they were the best potential state of affairs. x86’s remarkable flexibility is what has let it get so fast, not it’s underpinnings; it is cumbersome and ineffecient, carrying the legacy of the 1980’s with every program run on it. This is not what will take computing into the future. The future lies in RISC (PPC chips) EPIC (IA64), and as-yet unknown architectures. Notice that one of these is already in Apples’s playbook. I don’t see why they should deviate from that. A new supplier, maybe. A new specific chip – definitely. An old archicture, never.
The interesting thing about x86, is that a P4 is no x86. After you hit the instruction decoder, its a custom RISC core on the inside. x86 has gotten to the point where its something of a common language spoken by several completely different processors. There’s not much in the x86 architecture itself (as x86-64 proves) that is holding CPU designs back. As for a new architecture, I doubt it. Engineering a solid architecture is an extremely long, and expensive process. There isn’t the money, the engineering talent, or even the benifet. Take Sun, for example. For all their work since the late 80’s, they still can’t make a multi-thousand dollar UltraSparc III perform faster than a $400 Pentium 4. I’d like to see a nice RISC architecture out there as much as anyone (and I couldn’t care less about binary compatibility, thanks to Gentoo but unless its faster, cheaper, or both, there is just no good reason.
Eugenia/OSAlert: On a related Apple story, OSAlert reader Prognathus sents us a discussion link with a detailed description of Jaguar as a counter-productive version of MacOS, written by a MacOS 9.x advocate.
That pretty much says ALMOST everything I had against Aqua, except for two points, but never mind about those. And I’m not even a Mac zealot, or a Mac OS 9 user.
However, Aqua would score even more worse against NeXTStep’s UI, argubly the best of its time.
jbett: this is an unrealistic dream people, while Apple may have considered this alternative in the past, didn’t we just hear another article about IBM and Mac teaming together?
It isn’t unrealistic. Tell me why it is unrealistic. No Mac user I have met ever told me in detail why it is unrealistic without me proving them wrong.
On the other hand, on IBM and Mac teaming up, I would say it is as much a rumour as x86 Macs. This is because no other company after post-Jobs Apple ever showcased a top end processor that Apple end up using.
Plus, even if the GPUL (name?) turns out great, it doesn’t have the economic scale to follow Moore’s Law and therefore like the G4 would head straight into obvilion….
…. UNLESS IBM releases a commodity Linux machines using the processor, and try to get clone makers to make commodity Linux machines.
NeoWolf: this is an unrealistic dream people, while Apple may have considered this alternative in the past, didn’t we just hear another article about IBM and Mac teaming together?
They said “Pentium” crushing, not “Opteron” or “Athlon” crushing. Besides, Apple would most probably rebrand the processors like they did with PowerPC, and most people wouldn’t know the difference.
jbolden1517: If Apple were going to make a processor switch the easiest route is to some sort of modified Power4 chip.
Obviously, but Apple should look long term, would this be better? Sure short term IBM may have good processors, or Motorola maybe finally coming out with the G5, but look long term, my friend.
jbolden1517: If that fails the Itanium 2 makes a great deal more sense than the x86.[…]
There is no argument against Itanium in terms of pricing. But…
a) Itaniums take a lot more power than PowerPC or x86 processors. Why? Because it was built for high end (I mean HIGH END) workstations and middle end servers.
b) Itanium emits too much heat to make it a feasible choice for iMac/eMac, iBook and TiBook.
jbolden1517: Its not just that Steve Jobs would need to eat crow, he would need to back off from completely true claims about why x86 stinks.
He never made any claims on why x86 stink (like how bad its architecture is, or how bad being bogged down by legacy stuff is), just how slow Pentiums 4 is. And he is wrong currently on that (that’s why he doesn’t harp on that all that much anymore, compared to before).
jbolden1517: x86 with its terrible power consumption would hurt Apple laptops quite a bit in terms of battery life and the laptops are strong margin products.
Intel is releasing Banias that uses less power than the G4 in Q2 2003. Plus, current power consumption difference between Pentium 4-m and Athlon XP Mobile isn’t much different than the G4.
jbolden1517: On the powermacs going dual seems to be a very workable strategy.
Unfortunately, I could buy a dual Xeon or a dual Athlon MP at a much better price than PowerMacs, and probably get a much better deal at it.
Plus, going dual only benefits SMP/multithread apps. Most apps that are SMP capable only uses it for things like rendering, so it doesn’t *feel* faster.
jbolden1517: The Apple XServe can afford to underperform on CPU because its real aim is ease of administration
And how much ease of administration can it pull off when Windows .NET Server comes out? Besides, there doesn’t seem to be a market for products like these (rackmounts that are easy to administrator). People buying XServes now (I’m not sure about USA, but in Malaysia and Singapore) is to stream Quicktime media.
For a normal web server, for example, not only is the processor underperforming, the OS itself is underperforming (compared with Linux and FreeBSD on the same hardware).
yc: It is a very will know fact that Apple is working with IBM on future processors.
Since when did it become a fact?
yc: People would not be interested in a proprietary Intel box
Since when people buy Macs for its processors?
Cesar Cardoso: AMD needs to have a major coup against Intel
And wow, Apple is gonna help them? (Look at Apple’s market share my friend, AMD isn’t interested in that market).
Mike Bouma: 1) Many currently available PPC Macintosh software must then be ported over to the new x86 platform.
And:
Mike Bouma: Enter Tao’s Virtual Processor!
You prove my point before I can say it.
Ronald: Well, porting software from PowerPC to the Intel Architecture should involve not much more than recompiling!
For people using Code Warrior, it takes much more that that. Plus, to take out AltiVec code and add in SSE2 and HyperTransport code is long, and what about those assembly apps?
Harbinjer: These chips are both bery expensive.
Read jbolden1517 post.
Ralf.: Maybe Euqenia sould switch the name of this site to AppleSwitchToX86Cant.
If that’s the case, there is a bigger case for renaming OSAlert as LinuxNews or .NETvsJavaNews etc.
Ralf.: Wake up boys and girls – OS9 is dead.
So what if it is dead? Apple just threw away once of its closest guarded asset with OS X, I think everyone, especially Mac users, have the right to bitch over the stupid UI.
Ralf.: OS9 is old, ugly and has a bad OS design.
Old can be fixed without changing the UI. Ugly also can be fixed without changing the UI. And the design also can be changed without changing the UI.
Ralf.: In the MS/Intel world no one will flavor Win3.1 over WinXP.
First because
a) Windows 3.1 is so much older compared to Windows XP, than OS 9 compared to OS X.
b) Windows 3.1 UI is so much worse than Windows XP.
jbolden1517: In general with battery Mac laptops give about an extra 2 hours of battery life over PC laptops. That’s a huge difference.
if you rip off the battery in your Mac notebook and place it in your PC: guess what? The battery life would be almost the same. Where the heck did you think PCs cut their laptop cost on? Certainly not processors (they cost the same as G4s…).
Besides, I manage to get 6 hours of battery life out of this notebook – no standby, doing a mix of Office and Photoshop work.
Of course, the amount of hours would be much less if I used the standard battery pack they gave.
jbolden1517: GCC has been out for years for the Itanium line. Its not going to cost Apple much to improve GCC on Itanium.
The GGC processor for Itanium, IIRC, is worse than Intel’s sucky compiler. And it takes a lot of costs for Apple to make a compiler optimized for the processor.
jbolden1517: Heck Intel might absorb the cost as part of the deal (and they have lots of experience in writing pretty good compilers). I’d consider this a non issue.
Intel doesn’t care about Apple – they wouldn’t bring in much money to bother helping them make a compiler.
jbolden1517: This one I have to disagree. The x86 line is very complex. Chip speeds are hitting all sorts of quantum limits. I think getting to a 3ghz Itanium 2 is going to be much easier then getting a 10ghz Pentium 4.
Funny, Intel is closer making a 10GHz Pentium (certainly not Pentium 4) than a 3GHz Itanium. How much has Itanium increase in clockspeed since Itanium 1?
HereSince00: If he took $150M USD from Bill Gates, then he’ll do whatever it takes to keep Apple moving forward.
He took it as part of a legal settlement. Plus, Jobs never said anything that directly bashes Microsoft outside of the courtroom.
Jay: Getting a much faster bus would help alot, then DDR RAM would be of real help then.
The processor must support it in order to get the extra preformance. Motorola is planning this, but I bet they would come up with PC2100 ot PC2700 DDR right after Intel and AMD introduces stuff that uses DDR-II.
The Prophet: think about the culture death that would occur if Apple went to Intel and DRM.
I’m tired of you and DRM.
If Apple goes the Intel way, it would in NO way be forced to use LaGrande. LaGrande is a optional feature for chipset makers. Plus, LaGrande was built for purposes other than DRM (mostly to compete with IBM).
Besides, as for how DRM goes, you wouldn’t be using it UNLESS you are watching/hearing DRM media.
js: As for having Itaniums in Powermacs, I’d like to see the size of the battery!
PowerBooks uses bateries, but PowerMacs?
mazzaroth: Not in terms of current performance, but in terms of computer science, x86 is a dead architecture.
PPC on workstations is closer to dead than x86.
mazzaroth: carrying the legacy of the 1980’s with every program run on it
That legacy part of the processor wouldn’t even NEED to be touched by any Mac program, only legacy applications (which what kept x86 in power, among other reasons).
mazzaroth: The future lies in RISC (PPC chips) EPIC (IA64), and as-yet unknown architectures.
PowerPC is not pure RISC, it has some CISC features. Current and future x86 processors on the other hand is not pure CISC, as it has a lot of RISC features.
Besides, the only reason why EPIC is better than x86 is that it doesn’t have any legacy stuff, but if you are comparing with CISC, EPIC is much worse in the way you jugde architectures.
How many times are you people going to post the same story OVER AND OVER AND OVER again? For god’s sake stop beating the dead horse.
bolden1517: If Apple were going to make a processor switch the easiest route is to some sort of modified Power4 chip.
Obviously, but Apple should look long term, would this be better? Sure short term IBM may have good processors, or Motorola maybe finally coming out with the G5, but look long term, my friend.
That’s why I’m pushing for 64 bit and not 32. If they are going to make a processor change (other than G5 and possibly Power4) then they need to get a decade from the processor; x86 is already very long in the tooth. It should be around for a decade the problems with this chip are getting more and more serious.
jbolden1517: If that fails the Itanium 2 makes a great deal more sense than the x86.[…]
There is no argument against Itanium in terms of pricing. But…
a) Itaniums take a lot more power than PowerPC or x86 processors. Why? Because it was built for high end (I mean HIGH END) workstations and middle end servers.
b) Itanium emits too much heat to make it a feasible choice for iMac/eMac, iBook and TiBook.
I think in terms of the iMac/eMac the cache reduction should take care of the heat. On iBook/Powerbook the cache reduction will probably do it and if not Intel could build a lowerer power model. In any case the Itanium 2 has far less transistors than the pentium 4 (excluding cache) and so I don’t see this as a huge issue.
jbolden1517: Its not just that Steve Jobs would need to eat crow, he would need to back off from completely true claims about why x86 stinks.
He never made any claims on why x86 stink (like how bad its architecture is, or how bad being bogged down by legacy stuff is), just how slow Pentiums 4 is. And he is wrong currently on that (that’s why he doesn’t harp on that all that much anymore, compared to before).
Sure he has. He’s complained about the huge pipelines at this year’s mac world for example.
jbolden1517: x86 with its terrible power consumption would hurt Apple laptops quite a bit in terms of battery life and the laptops are strong margin products.
Intel is releasing Banias that uses less power than the G4 in Q2 2003. Plus, current power consumption difference between Pentium 4-m and Athlon XP Mobile isn’t much different than the G4.
Its quite possible new work tilts the balance. I can only go on what exists. But the fact is there is no good reason the Pentium 4 shouldn’t use the more power and the more heat of any of this list (with normalized cache): G4, Power4, Itanium 2…
jbolden1517: On the powermacs going dual seems to be a very workable strategy.
Unfortunately, I could buy a dual Xeon or a dual Athlon MP at a much better price than PowerMacs, and probably get a much better deal at it.
I agree but it isn’t the chips that’s causing that; it’s Apple’s higher margins, changing chips won’t fix that problem (at least not to Apple’s satisfaction).
Plus, going dual only benefits SMP/multithread apps. Most apps that are SMP capable only uses it for things like rendering, so it doesn’t *feel* faster.
Actually here I disagree. Going dual is wonderful for single threaded CPU intensive tasks since the stays responsive even under heavy loads. Its much more comfortable using 1 out 2 CPUs for 3 hours then 1 out of 1 CPUs for 1 hour if you need to use the box for anything else at the same time. Of course on multi threaded apps you get the drop of responsiveness but almost double in speed.
jbolden1517: The Apple XServe can afford to underperform on CPU because its real aim is ease of administration
And how much ease of administration can it pull off when Windows .NET Server comes out?
I see no evidence that Microsoft is designing .NET server to be much easier than the NT/2000 Server line. Apple is the king of ease of use I think they’ve established some credibility here.
Besides, there doesn’t seem to be a market for products like these (rackmounts that are easy to administrator). People buying XServes now (I’m not sure about USA, but in Malaysia and Singapore) is to stream Quicktime media.
For a normal web server, for example, not only is the processor underperforming, the OS itself is underperforming (compared with Linux and FreeBSD on the same hardware).
I think the difference is the cost of labor in the USA. One skilled Unix admin costs 12-30 XServes per year. For small business or offsite locations if you can avoid having to hire highly skilled admins by spending more on hardware its a no brainer.
jbolden1517: In general with battery Mac laptops give about an extra 2 hours of battery life over PC laptops. That’s a huge difference.
if you rip off the battery in your Mac notebook and place it in your PC: guess what? The battery life would be almost the same. Where the heck did you think PCs cut their laptop cost on? Certainly not processors (they cost the same as G4s…).
Besides, I manage to get 6 hours of battery life out of this notebook – no standby, doing a mix of Office and Photoshop work.
Of course, the amount of hours would be much less if I used the standard battery pack they gave.
I have an Inspron 8000 which was Dell’s top of the line. Battery life is 2-3 hrs per battery under normal usage. The laptop at the time was over $4k so cutting costs on the battery doesn’t make much sense. To get up to Apple’s lifetime I have to go dual battery. I’d love to be able to get over 8 hrs going dual.
jbolden1517: GCC has been out for years for the Itanium line. Its not going to cost Apple much to improve GCC on Itanium.
The GGC processor for Itanium, IIRC, is worse than Intel’s sucky compiler. And it takes a lot of costs for Apple to make a compiler optimized for the processor.
jbolden1517: Heck Intel might absorb the cost as part of the deal (and they have lots of experience in writing pretty good compilers). I’d consider this a non issue.
Intel doesn’t care about Apple – they wouldn’t bring in much money to bother helping them make a compiler.
Intel already has the compiler all they have to do is make it available to GCC. Its not like Intel makes a lot of money of their compilers. Improving GCC also helps the BSD and Linux crowd. Intel writes compilers for the purpose of selling CPUs.
Also remember Itanium is Intel’s new baby; they want this chip in mainstream products. Apple is certainly mainstream in a way that BSD and Linux are not.
Finally Intel would care about Apple for the same reason they care about Linux; it gives them some leverage over Microsoft.
jbolden1517: This one I have to disagree. The x86 line is very complex. Chip speeds are hitting all sorts of quantum limits. I think getting to a 3ghz Itanium 2 is going to be much easier then getting a 10ghz Pentium 4.
Funny, Intel is closer making a 10GHz Pentium (certainly not Pentium 4) than a 3GHz Itanium. How much has Itanium increase in clockspeed since Itanium 1?
Good point. I have no idea why Intel isn’t focusing more on speed. I’ll stand by my statement but you did score there.
It’s cool!
I tend to agree with the basic idea that a chipmaker would not make a chip specifically for Apple, based on the idea that Apple only has 2-4% of marker share (depending on which stats you look at). On the other hand, looked at in a different way, Apple does sell millions of computers each year. A chipmaker would find it worthwhile to at least perhaps modify a chip *if* the bottom line meant profit. To me, considering Apple’s corporate/cultural history, the variation of the Power4 would be the best candidate. And, as I said above, they have some time, as they can string out the G4 if they do certain things. In particular, Apple has to get a modern, much faster bus, no matter what else they do. They have to. If they don’t do that then faster processors will mean almost nothing. So, it seems to me they would work on that while stringing out the G4. And then possibly start making some decisions.
Everyone know mac is going to go with the GPUL from ibm i mean if you dont you just dont know what you are talking about. My apple rep where i work knows a few higher up guys in apple to and he says just about everyone their is sure apple is going GPUL and never even really considered x86 seriously
Why doesn’t anyone believe that the x86 is a (blank) toy? There are so many design problems that have been patched and patched over and over again that it is just plain stupid to waste time on such junk!
Oh great. And if the G4 can’t go above 1.2 GHz what do u think it is, if not a chip design problem ?
The real point is, despite all the design problems you may enumerate, top-of-the-line P4’s and Athlon’s are much, much, faster than top-of-the-line G4’s.
Don’t tell me it went to the moon again please. That was a limited and HIGHLY tested single use that didn’t use DOS.
So ? Windows 2000/XP, Linux, *BSD, don’t use DOS either. On the other hand, MacOSX still has an emulation layer for a non-preemptive multitasking system, hehe ;-))
apple may have 2-4% of the marketshare, but they are selling over 10% of the new systems.
a new processor company would be getting over 10% of the total new desktop cpu market, not 3%. And considering a current 3% market but some 10-12% sales of new macs obviously in a few years total market share is going to go up.
Intel already has the compiler all they have to do is make it available to GCC.
All they have to do… You obviously don’t understand the problems. The problem is that GCC’s current internal architecture is not adapted to optimizing code for EPIC CPU’s, or even RISC. So you would have to rewrite all of GCC, while continuing to have support for tens of different architectures (GCC is a multi-platform & multi-processor compiler, case you don’t know).
Apple shoud license the alpha-core. I think it’s quiet cheap these days . And it’s one of the best ( if not the best ) architecture availabel today.
rajan r:
> Cesar Cardoso: AMD needs to have a major coup against Intel
> And wow, Apple is gonna help them? (Look at Apple’s market > share my friend, AMD isn’t interested in that market).
The problem of AMD nowadays is NOT market share, is mindshare.
Apple has LOTS of mindshare, even if their market share remains tiny.
Apple+AMD = (ideally) more mindshare for AMD and more market share for Apple.
Ralf:
> OS9 is old, ugly and has a bad OS design.
Take out the dock and ALL the fruity waste of CPU cycles, and what OS X is REALLY different of OS 9 in terms of GUI?
> It should be around for a decade the problems with this chip are getting
> more and more serious.
x86 is not a chip; it is an architecture. The architecture does not have “serious problems.”
> Sure he has. He’s complained about the huge pipelines at this year’s mac
> world for example.
Those “huge” pipelines allow the P4 to hit much higher clock speeds than the G4 and thereby outperform it.
> Its quite possible new work tilts the balance. I can only go on what
> exists. But the fact is there is no good reason the Pentium 4 shouldn’t
> use the more power and the more heat of any of this list (with normalized cache):
Since when does theory trump reality?
> Of course on multi threaded apps you get the drop of responsiveness but
> almost double in speed.
Most multithreaded applications do not gain that amount of speed. It is more like a 30-50% increase. The only reason to get a dual-processor machine is when the dual-processor machine outperforms a uniprocessor machine at the same price level. Or in other words, for the vast majority of desktop users, getting a 2 GHz machine would be a far better choice than a dual 1.2 GHz machine. The 2 GHz G4 does not exist in a stable, manufacturable form, so Apple sells the dual 1.2 GHz machine.
> I have no idea why Intel isn’t focusing more on speed.
Er…how does focusing on the P4 mean that Intel is not focusing on speed? The P4 is the fastest desktop chip available at 2.8 GHz, and even at the new speed of 3.33 GHz it will be operating at one-third of its expected maximum speed. Intel plans to take the chip up to 10 GHz with new manufacturing methods.
How would scapping the P4 production lines to produce a new CPU based on an architecture for which compiler writers find extremely hard to optimize code actually be “focusing more on speed”? 64-bit does not mean double the speed; 64-bit means significantly increased addressing space. I have yet to see a great need for a desktop computer with more than 64 GB of RAM, yet that is the limit of the 36-bit addressing on the x86 chips.
Tell me, why do we need 64-bit chips?
> On the other hand, MacOSX still has an emulation layer for a non-preemptive
> multitasking system, hehe ;-))
So does Windows XP for both Win16 applications and DOS. But that does not necessarily mean that MacOS X or Windows XP are bad.
The x86 decode unit is extremely small these days compared to the rest of the cpu. Way less than the number of transistor difference between p4 and athlon.. and if u remember correctly the x86 decodeing certainly dosent require 10 stage pipeline. Thats for performance.
Yes pipelines suck, but they enable u to clock it higher.. the reason a p3 can beat a p4 isnt the pipeline (just) its mainly just the redesign of the chip for SSE2. Many instructions take longer so that a few can be faster.. This is the organisation and bottleneck management of the core cpu pieces .. not a x86 decode latency or anything like such.
Removing x86 from x86 chips will not gain any speed, by the time it was prefabbed it would allready be behind the latest tweaked version which hasnt been stalled for the point of removing something that directly wont bring any benefit from being removed.
The Real performace bottlenecks.. are not x86 decode thats for sure. (Btw the perf bottle necks are mem bandwidth (hello g4 and ath to a much much lesser extent) and cache misses (hello Itanium 2 with it compiler hinting;).. in complex code pipeline stalls happen 2.. complex like 3dsmax rendering.. which the p4 is catching up if not beating the athlon (p4 more pipes more SSE2 optimised) nowdays. Mpeg4 rendering the p4 cains.
P4 might be relativly slow at non optimsied scientific and compiling code.. but it runs SSE2 real fast and so far no athlon or g4 can even do 64 bit percision SIMD. Perhaps we shouldnt piss on the fastest proc.. that goes up to now 4.7 gig demoed by intel.
Just perhaps
I read that somewhere in these long posts that someone actually thinks you get the same batterylife in aPC notbook as an Apple notebook. R U NUTZ!!!! I have an ibook and a buddy just bought a new Thinkpad. they are aboutthe same size He gets 1/2 the batterylife. Dont get me wrong his is MUCH faster but still if you are talking batterylife get real.
As far as PPC vs any ATHLON or PENTIUM, the PPC is a new chip as far as development goes. I do think that APPLE really has to consider all options. IBM’s new chip from what I read wont be ready for about a year. Either Apple has something to get more life out of the G4 line to wait for the IBM chip or it is ATHLON time if I had my way.
And what is this about Aqua I miss some aspects of OS9 but it is not the interface. The unix/apache directories and files take some getting used to but that is about it. I love OSX even over my XP machine.
Thinking here, yes, it would be suicide to market Mac OS X for Intel, but how about OS X Server? Couldn’t a five houndred dollar price tag pay for a porting effort?
Apple could offer it only through OEMs, so they wouldn’t have to support every crap video card ever designed. Also, buyers would have a better idea what they are getting into since “Joe User” wouldn’t be lineing up.
The “Steve’s Ego” factor might still be a problem (this is the guy who left Bill Gates waiting around the Next office until he finally left), but I’m sure there is some way out of that.
Note: I have not read past the first fifteen posts, so someone may have already brought this up. Also, I don’t watch southpark, and can’t remember the word, so I’m not going to refer to Malarkar.
And what is this about Aqua I miss some aspects of OS9 but it is not the interface. The unix/apache directories and files take some getting used to but that is about it. I love OSX even over my XP machine.
The article doesn’t compare Aqua and Luna, but Platinum and Aqua.
jbolden1517: If they are going to make a processor change (other than G5 and possibly Power4) then they need to get a decade from the processor; x86 is already very long in the tooth.
One word: x86-64 (http://www.x86-64.org/). The problems you mentioned with x86 just now is problems with the chip, the processor. Problems that would be fixed in less than a year.
jbolden1517: I think in terms of the iMac/eMac the cache reduction should take care of the heat. On iBook/Powerbook the cache reduction will probably do it and if not Intel could build a lowerer power model.
Cache reduction doesn’t solve (though help) the heat problems, and it certainly does solve the power problems. And like I said earlier, it is very unlikely that Intel would team up with Apple in promoting two competing products in the same market.
With this in mind, Apple is better off sticking on Intel’s IA32 line until 64GB RAM becomes a real problem, whom a switch to AMD x86-64 would be very painless.
jbolden1517: In any case the Itanium 2 has far less transistors than the pentium 4 (excluding cache) and so I don’t see this as a huge issue.
But the motherboard required to run it uses a thousand times more transistors than a high end P4 motherboard.
jbolden1517: Sure he has. He’s complained about the huge pipelines at this year’s mac world for example.
If you call that complaining….. I don’t know what to call complaining anymore.
jbolden1517: Its quite possible new work tilts the balance. I can only go on what exists.
And unlike G5s and Power4s and GPULs, Banias have been showcased, computer manufacturers had announce they are supporting it etc.
While it doesn’t exist now, Apple isn’t switching tommorrow, they would be doing it after they are satisfied with the transition to OS X.
jbolden1517: But the fact is there is no good reason the Pentium 4 shouldn’t use the more power and the more heat of any of this list (with normalized cache): G4, Power4, Itanium 2…
Pentium 4 uses much less power than Itanium 2 with the same amount of cache (can’t compare with Power4 cause it doesn’t have any version with that little cache). G4 may be more power efficient NOW, this would probably change soon.
jbolden1517: I agree but it isn’t the chips that’s causing that; it’s Apple’s higher margins, changing chips won’t fix that problem (at least not to Apple’s satisfaction).
My point wasn’t in the price (the price doesn’t matter for Apple’s target market…. well, almost). But rather performance.
jbolden1517: Going dual is wonderful for single threaded CPU intensive tasks since the stays responsive even under heavy loads.
I notice no performance difference between a 867MHz G4 (dual) and an old one running at similar clockspeeds when it comes to non-multithreaded apps. Plus, they aren’t more responsive.
jbolden1517: I see no evidence that Microsoft is designing .NET server to be much easier than the NT/2000 Server line. Apple is the king of ease of use I think they’ve established some credibility here.
From all the people I know that runs Windows .NET Server, it is much easier to administrator than Windows 2000 (none of them, however, administrator NT4 servers before, so they and I wouldn’t know).
jbolden1517: For small business or offsite locations if you can avoid having to hire highly skilled admins by spending more on hardware its a no brainer.
Trust me, no sane company would adopt a platform and hire some teenager in the street to administrator it because it is easy to administrate. It is the same reason why many companies still hire MSCE admins to run their servers even though the owner have the time to do it himself, and could do it himself.
jbolden1517: I have an Inspron 8000 which was Dell’s top of the line.
I wouldn’t recommend any Dell laptop to anyone. Especially that laptop. If you want a power-efficient laptop that performs roughly the same as its Apple counterparts, you picked the wrong laptop, my friend.
That very model you choosed is for people you want their desktops mobile (e.g. same power as a desktop).
jbolden1517: Intel already has the compiler all they have to do is make it available to GCC. Its not like Intel makes a lot of money of their compilers.
It may be small compared to their other products, but it is still a money maker. Plus, Itanium’s compiler is significantly different from GCC, making it easier to rewrite GCC than to merge them.
jbolden1517: Also remember Itanium is Intel’s new baby; they want this chip in mainstream products.
No, they never made this processor for mainstream. It may trickle down after a couple of decades or so, but itanium was made to get into a profitable niche in which Sun currently dominates.
jbolden1517: Finally Intel would care about Apple for the same reason they care about Linux; it gives them some leverage over Microsoft.
They don’t care about their leverage on Microsoft. They supported Linux because it was vital. It is a growing player, one that is almost dominate in the server market. Apple on the other hand isn’t growing market share wise.
jbolden1517: Good point. I have no idea why Intel isn’t focusing more on speed. I’ll stand by my statement but you did score there.
Because they target market isn’t interested in clock speed, plus Itanium’s architecture in general makes it difficult to add in clockcycles.
Cesar Cardoso: The problem of AMD nowadays is NOT market share, is mindshare.
AMD has much more mindshare than Apple, much more brand recogniction and much more press than Apple ever had. Apple may have a good PR and brand for a company that size, but we are talking about AMD here. AMD doesn’t need Apple mindshare, trust me.
You wanna know how good is AMD’s mindshare? Go to any randomly picked PC geek forum/message board, and post something anti-AMD, and be sure to pray that you would get out alive.
Cesar Cardoso: Take out the dock and ALL the fruity waste of CPU cycles, and what OS X is REALLY different of OS 9 in terms of GUI?
OS X’s deefiencies in Aqua goes MUCH deeper than that. Simple because it is a mish mash of two completely different UIs.
How much battery life do iBooks and TiBooks get? My Compaq 2800T, which is a 1.8GHz P4 with a 15″ UXGA screen and 64MB Radeon 7500 gets 3.5 hours even when on wireless.
Hahah, with it’s screensaver running and the hard disk spun down I might believe you.
My TiBook gets me between 3 – 4 hours, my Dell Inspiron about 2 – 2 1/2
Hell-lo-ho!
#1 power consumption in a laptop is not the CPU, but the display.
And you get what you pay for. My Gateway latop runs 4 hours straight with the stock battery – as long as I don’t use the CD drive, which cuts battery life to ~30 minutes.
Please stop using battery life as a benchmark for CPU efficiency. That’s like using RC5 for performance benchmarks…
AC, notice that the Compaq uses a Radeon, and a display larger than the TiBook, and this is probably what drains the battery life.
I agree with what Solar said, battery life test is not a benchmark on how good a processor is. I remember reading a comparison of Pentium III-m laptops when they started coming out. The difference in battery life ranges by 3 hours! Yes, THREE HOURS.
There are other power draining components in a laptop other than a CPU.
The New York Times has an article today about how PC sales are flat because users are finding that further hardware speed increases don’t make software any faster. ( http://www.nytimes.com/2002/09/30/technology/30SPEE.html )
Is it really worth it for Apple to go to all of the horrendous trouble of a platform switch for a fractional speed increase?
I was not comparing speeds I was saying that put a PC and PPC laptop generally whith similair size screen and battery ect the PPC will last longer. I have yet to see a real world PC laptop have longer batterylife, I have seen some with bigger heavier batteries and that is fine and cool a extra 1lb or so is worth the extra life right.
Back to either PPC or X86 chip. I am all for either now. I like the PPC since I believe if they invested even 1/4 of what INTEL puts into its chips that the PPC would be a great fast efficient trip. Problem is that is a pipe dream. So I think either the IBM chip or X86. I am hoping that the IBM chip is going to be great but who knows.
Then again the new 64bit AMD chips are looking SWEET.
Joe user, speed improvements won’t, I repeat, WON’T be felt in UI resposiveness. Extra speed is used for stuff like extra FPS for gamers, faster rendering times for stuff like AfterEffects and so on.
Plus, most applications on Windows do NOT optimize for SSE or even 3DNow, so what could you expect?
So, if you aree just web browsing, or writing a school report, 2.8GHz P4 won’t really help you. But if you are that extreme gamer, where every FPS lost is like loosing a child, go ahead.
But for Macs, on the other hand, because of its UI, they are better off with x86. The last time I checked, scrolling and resizing Mac OS X’s apps isn’t the fastest thing…