Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Did Apple Do The Right Thing in Moving To Intel From PowerPC???
- Thread starter rashdown_online
- Start date
- Sort by reaction score
I admit I don't know the details of the programming but I do distinctly remember Steve and several pubs after the intel announcement was made, stating the UB coding allows Apple to move between Intel and IBM. If my memory serves me, I even recall some mentioning how Apple may want to move into the high end server market with IBM and would still be able to use OSX due to the UB nature.
That being said, Apple will NOT use the Power6 for any of its computers. For those of us who have been around before powerpc even existed we remember when the Power 4 was announced and it took longer to get a PowerPC chip. Not to mention how long we waited for updated chips before the Intel move and that NEVER materialized, which I think were supposed to be based on the Power5 chips.
IF so, then, why didn't Apple keep using PPC on workstations? Universal Binaries would have made it possible. Instead, they chose to abandon PPC completely for now at least.
Because it would be dumb to maintain two totally different CPU-architectures, that's why. And maybe because they can get better performance with Intel than they can get with PPC?
You can thank AMD for that. That dual independent bus" in G5? It's called HyperTransport. Invented by AMD, and it made it's debut in their Opteron-processor.
Seriously: are you guys totally forgetting the fact that before Intel came out with the Core-CPU's, they were getting spanked by AMD for years?
I miss the PowerPC
I miss PowerPC... Hardware updates are so boring now... Sure, Intel is the standard-- that's the whole problem. "Oh look, we've been updated to baseline performance again". The only suspense is speculating when Moore's Law will break down-- other than that, any pundit can tell you how fast your computer will be next year and the year after that. I miss looking forward to the next leap ahead.
Technically, the x86 is just an ugly architecture. They've built a superscalar structure that looks like the Winchester mystery house. PowerPC is a much more elegant design, and it should be because it's newer. It's not about RISC vs CISC-- those lines are gone. It's because PowerPC carries less cruft.
PowerPC is cheaper and lower power. When it's updated, it's faster than the current generation of x86 but unfortunately they aren't updated as often.
From a business perspective, Apple reduced their uncertainty by going to Intel, and they gave a bridge to switchers. It was probably a good business move, and I am really impressed with how smoothly the transition has been pulled off. From a technical perspective, I'm kinda grumpy about the whole thing (you've probably noticed that...).
Incidentally, I think it's too early to tell whether Apple undercut their business by giving developers an out and letting Windows apps be "cross platform".
I'd love to see Apple keep both lines running in parallel-- POWER, or PowerPC, in XServe and the mini, Intel on the desktop, and which ever architecture is leading the power race in the portables. I love my G4 mini, and I'd love to pick up a couple more cheap mini's. Trying to build an inexpensive product around a $400 chip just seems silly.
CPU options, coupled with Universal Binaries would really give Apple an edge in the market, in my opinion. For this to work though, they'd have to get another PPC product out before developers stop making UBs in favor of Intel only builds. They'd also have to convince the market that this was the best of both worlds rather than vacillating.
Intel is not the R&D leader in CPUs. IBM's cell processor is much more cutting edge than anything Intel has put out. Core2 benefited more from Alpha than Intel's internal developments. Intel has built a brutal, cut throat, and successful business, but if you're going to credit their R&D give your accolades to the process engineers, not the design group.
I miss PowerPC... Hardware updates are so boring now... Sure, Intel is the standard-- that's the whole problem. "Oh look, we've been updated to baseline performance again". The only suspense is speculating when Moore's Law will break down-- other than that, any pundit can tell you how fast your computer will be next year and the year after that. I miss looking forward to the next leap ahead.
Technically, the x86 is just an ugly architecture. They've built a superscalar structure that looks like the Winchester mystery house. PowerPC is a much more elegant design, and it should be because it's newer. It's not about RISC vs CISC-- those lines are gone. It's because PowerPC carries less cruft.
PowerPC is cheaper and lower power. When it's updated, it's faster than the current generation of x86 but unfortunately they aren't updated as often.
From a business perspective, Apple reduced their uncertainty by going to Intel, and they gave a bridge to switchers. It was probably a good business move, and I am really impressed with how smoothly the transition has been pulled off. From a technical perspective, I'm kinda grumpy about the whole thing (you've probably noticed that...).
Incidentally, I think it's too early to tell whether Apple undercut their business by giving developers an out and letting Windows apps be "cross platform".
I'd love to see Apple keep both lines running in parallel-- POWER, or PowerPC, in XServe and the mini, Intel on the desktop, and which ever architecture is leading the power race in the portables. I love my G4 mini, and I'd love to pick up a couple more cheap mini's. Trying to build an inexpensive product around a $400 chip just seems silly.
CPU options, coupled with Universal Binaries would really give Apple an edge in the market, in my opinion. For this to work though, they'd have to get another PPC product out before developers stop making UBs in favor of Intel only builds. They'd also have to convince the market that this was the best of both worlds rather than vacillating.
Here lies the crux of the misperceptions. A successful business doesn't indicate superior products.
Intel is not the R&D leader in CPUs. IBM's cell processor is much more cutting edge than anything Intel has put out. Core2 benefited more from Alpha than Intel's internal developments. Intel has built a brutal, cut throat, and successful business, but if you're going to credit their R&D give your accolades to the process engineers, not the design group.
Very good point (though I think the Xserve is distinct as well given the fan noise in that beast)...
As a company, The Intel transition has been the best thing since System 7. The might not have the most powerful processor in the world available, but their Market share has soared like never before to become competitive in the computer market again. When Boot Camp was released, suddenly Apple became an viable option for anyone unsure about the Apple platform, and they were a serious market competitor again.
The next revision of the MacBook Pros after Boot Camp release sold like Hot Cakes (Too bad that the Drivers suck, and the original Intel Macs were not designed with Windows support in mind)
Of course the options are still there to open up Mac OS X 10.5 PPC to be licensed by third party server vendors for use on the POWER6, or even make a quiet release of a POWER6-based server (As not to upset Intel)
The next revision of the MacBook Pros after Boot Camp release sold like Hot Cakes (Too bad that the Drivers suck, and the original Intel Macs were not designed with Windows support in mind)
Of course the options are still there to open up Mac OS X 10.5 PPC to be licensed by third party server vendors for use on the POWER6, or even make a quiet release of a POWER6-based server (As not to upset Intel)
They *are* maintaining two CPU architechtures, because of the Universal Binaries! It's not that they have *abandoned* PPC code, it's just that they have *embraced* Intel code. All the pieces for new PPC hardware *are* there, and software-wise Apple can release new PPC hardware anytime they want to.
That's a better explanation; however, it's up to IBM mostly. If they put out a new super chip faster than anything Intel has to offer, then would Apple not use it? Competition.
Well judging by this poll a huge gigantic enormous majority of Mac users think so.
Yes!!!
Yes!!!
Before abandoning the Pentium architechture in favor or Israeli-designed Pentium-M, they were spanked by everybody. Even Motorola could keep up with the absolute performance of Pentium chips, and that's not much. Pentiums wasted too much energy to heat up the whole house instead of putting it into number crunching. Embracing the performance-per-watt ideology of Pentium-M actually *saved* Intel.
Er ... dog show singular? Translation please as it's not in my Oxford Dictionary.
they are doing that not by choice, but because of necessity. Do you think that Apple LIKES maintaining two difference architectures?? And now you are suggesting that they should not only continue to do that indefinitely, they should also reinforce it? And for what? Because they MIGHT get MARGINALLY faster CPU's from IBM than they could get from Intel?
Sooner Apple can rid themselves of that codebase, the better off they are. But since they have large install-base of PPC-hardware, they have to support it. Just because they have to do it, does not mean that they like doing it. And it makes ZERO sense to suddenly start releasing PPC-hardware.
No, because it would not be smart in the big picture. It would just confuse the marketplace and developers. It would also diminish the economies of scale.
So should Apple release POWER6 (or it's derivative)-based systems? What if one month later Intel released a CPU that was faster than POWER6, should Apple then replace POWER with Xeons? And then replace Xeon with POWER when IBM leapfrogs Intel? No, that does not make any sense, and it would just lead to madness. You take one processor-architecture and stick with it. You do not go around switching architectures just because the "other" architecture happens to be somewhat faster at this very moment.
Slow dev from IBM, low priority
As has been pointed out, the G6 would have been a stripped down version of the power 5, or if the G6 was based on the power 6 we could expect G6 macs some time in 2009-2010. The reason that Apple was forced to abandon IBM processors was because IBM could not continue to develop them when they were produced for a tiny production run and were probably about IBMs 107th priority (the chips coming from Intel are their first and foremost priority). The mobile G5 was released 3 years after the desktop chip, which you have to remember was not fun for people running 6 year old architecture in their last rev G4 powerbooks. Apple was buying product in orders in the hundred thousands, and selling probably 2 million processors per year. IBM was more committed to chips for Nintendo, Sony, and Microsoft based on power than Apples needs, so they would have had to make computers based on the Xbox cpu branded as G6 more than likely, or possibly Cell. Going with an off the shelf general purpose chip rather than a purpose designed console chip was a no-brainer. I was horrified when they announced this amid the death throws of the Pentium IV, which was plagued by the worst design choices in Intel history. With what they have done since, after having the snot beat out of them by AMD, the transformation of Intels product line and consequently the Mac has been fantastic.
Jobs was right.
As has been pointed out, the G6 would have been a stripped down version of the power 5, or if the G6 was based on the power 6 we could expect G6 macs some time in 2009-2010. The reason that Apple was forced to abandon IBM processors was because IBM could not continue to develop them when they were produced for a tiny production run and were probably about IBMs 107th priority (the chips coming from Intel are their first and foremost priority). The mobile G5 was released 3 years after the desktop chip, which you have to remember was not fun for people running 6 year old architecture in their last rev G4 powerbooks. Apple was buying product in orders in the hundred thousands, and selling probably 2 million processors per year. IBM was more committed to chips for Nintendo, Sony, and Microsoft based on power than Apples needs, so they would have had to make computers based on the Xbox cpu branded as G6 more than likely, or possibly Cell. Going with an off the shelf general purpose chip rather than a purpose designed console chip was a no-brainer. I was horrified when they announced this amid the death throws of the Pentium IV, which was plagued by the worst design choices in Intel history. With what they have done since, after having the snot beat out of them by AMD, the transformation of Intels product line and consequently the Mac has been fantastic.
Jobs was right.
Yes. All the x86-CPU's are basically RISC-CPU's. They have a thin layer that converts the CISC-insctructions in to RISC-instructions, and they are then internally handled as RISC.
But POWER6 is A LOT faster
It's also A LOT more expensive. Seriously guys, we are not talking about a CPU that costs few hundred bucks, we are talking about a CPU that costs THOUSANDS of dollars.
But POWER6 is A LOT faster
But how many consumer SERIOUSLY would use 10% of the power and features with a POWER6?
Office, Safari and iTunes would start up pretty fast
In Intel-land, the main competitor to POWER6 is Itanium 2. Wikipedia gives Itanium 2 these launch-prices:
* Itanium 2 9050 1.60 GHz / 24 MB L3 -- $3,692
* Itanium 2 9040 1.60 GHz / 18 MB L3 -- $1,980
* Itanium 2 9030 1.60 GHz / 8 MB L3 -- $1,552
* Itanium 2 9020 1.42 GHz / 12 MB L3 -- $910
* Itanium 2 9015 1.40 GHz / 12 MB L3 -- $749
* Itanium 2 9010 1.60 GHz / 6 MB L3 / single core --$696
That should give you some idea to the pricing of POWER6 (assuming it was available to companies outside IBM).
* Itanium 2 9050 1.60 GHz / 24 MB L3 -- $3,692
* Itanium 2 9040 1.60 GHz / 18 MB L3 -- $1,980
* Itanium 2 9030 1.60 GHz / 8 MB L3 -- $1,552
* Itanium 2 9020 1.42 GHz / 12 MB L3 -- $910
* Itanium 2 9015 1.40 GHz / 12 MB L3 -- $749
* Itanium 2 9010 1.60 GHz / 6 MB L3 / single core --$696
That should give you some idea to the pricing of POWER6 (assuming it was available to companies outside IBM).
Yes, exactly. Apple likes to have a choice on what hardware to use. Because of having Universal Binaries, OSX and UB-era Mac software are CPU-architechture-agnostic which means Intel or any other CPU manufacturer cannot bully on Apple. In fact, take a look at IBM they tried to not do their part of the contract (in developing G5 further) and Apple didn't look at it very long. Because they had had in-house Universal Binaries of OSX for years, it was a piece of cake for them to release Intel-based version. The hard part was to get 3rd party software houses to do the same, but now that most major software titles are Universal, it gives Apple even greater freedom to choose the platform(s) that make the best sense. If a new architechture has to be supported, it's a matter or re-compiling for Apple and UB-compatible 3rd parties.
It takes at least 3 years for Apple to get rid of support contracts of PPC hardware. In 3 years storage space has grown so much it makes zero sense to drop support for the architechture that can be usable for 10 years more. If nothing else, Apple can justify selling that bigger storage as the software releases keep growing bigger over time.
Because the support is and will be there Apple has zero reason *not* to release PPC hardware if it would make sense for them and/or customers. If IBM for example would come out with something insanely great, Apple could be using it in a heartbeat.
I'm not talking about quarterly benchmarks, but if it happens to be so in the long term, Apple has zero reason not to switch again. It's called "competition" and it's only great that Apple has succeeded to position itself to be able to choose from different architechtures which can compete against each other directly.
Take Microsoft for example, they're stuck with x86 as there's no software support for other platforms. There used to be Alpha and PPC support for Windows NT, but Microsoft dropped it because they could not get 3rd parties to support multiple platforms. This is the key point, actually. Apple has 3rd parties in the multi-platform wagon and for the first time in computing history there actually is a meaninful choice!
No, Apple will not drop PPC support. As long as compilers support it, all 3rd parties will support PPC as well.
If Intel starts to bully Apple, they can always move to AMD. And it IS hard to support two totally different architectures. Yes, Apple said that all it takes is one checkbox, but in reality it is quite a bit harder than that. And just try to imagine the reaction when Apple first tells everybody that "we are done with Power and we are moving to Intel", and then a bit later saying "we are going back to PPC". It would make Apple seem confused and without direction.
Was it "piece of cake" for third-party developers as well? Was it "piece of cake" for Apple to _maintain_ those parallel-builds?
And it still doesn't make any sense to switch and then re-switch architectures.
Sure it does. They might think of exciting new ways of doing things, but they couldn't do it because "it doesn't work on PPC". Should they hold back their developement because of PPC? Like it or not, supporting PPC IS an extra drain on their resources that could be used elsewhere.
And it doesn't make sense for Apple nor it's customers. So why are we having this dicussion?
they could, but they wouldn't.
They have lots of reasons to not switch. Developers would get sick and tired of Apple jumping all over the map. Apple would lose face by annoucing the switch to Intel and then suddenly re-switching back to PPC, the support for several separate architectures would be a drain on their resources, it would be confusing to their customers (you can run Windows on this machine, but not on this machine")
And they are already reaping the benefits of that. If Intel screws them over, there's always AMD.
Seriously, you are advocating something that is completely unrealistic and dumb. You are basically saying that Apple should switch back to PPC because IBM released a hi-end CPU that happens to be very fast indeed. And how exactly is that hi-end CPU relevant to Apple AT ALL? The CPU Apple would use would NOT be this benchmark-shattering POWER6 we have been talking about. It would be slower and cheaper derivative of it. Would THAT version be competetive with CPU's from Intel?
This assumption is the flaw in your argument.
Fat binaries are *not* free to the application vendors. They are forced to build with Apple tools (something that is a very real pain point for cross-platform applications), they have significantly higher development and Quality Assurance costs, and the support costs are higher.
PPC will die out quickly, especially when you start to see that many of the 64-bit applications for 10.5 are for Intel 64-bit only.
Fat binaries will stick around, but only for the Intel apps that need to ship both 32-bit and 64-bit versions.
I would be surprised if OSX 10.6 even supports PPC. In 3 years, PPC won't be important.
Yeah, forgot about the Xserve.
Maybe I should have said all CONSUMER Macs are based on a laptop platform.