[Eug]

IBM Unleashes 4.7 GHz POWER6 Microprocessor

The POWER6 chip has a total cache size of 8 MB per chip – four times the POWER5 chip – to keep pace with the processor bandwidth. With 300 GB/s on tap, IBM boasts that its processor has so much bandwidth that the POWER6 chip could download the entire iTunes catalog in about 60 seconds.

To facilitate the lower energy demands of the new chip, the POWER6 designers separated circuits that can’t support low voltage operation onto their own power supply “rails,” allowing IBM to dramatically reduce power for the rest of the chip. IBM engineers also used new method of chip design that enables POWER6 to operate at low voltages, allowing the same chip to be used in low power blade environments as well as large, high-performance symmetric multiprocessing machines.

[Dakarʒ]

Is this what could have been considered the G6 or something?

[mdc]

I wonder with Universal Binaries if Apple would ever make a Mac Pro, or iMac with another PowerPC chip in it? Or are they Intel all the way now.

[Person Man]

Originally Posted by Dakarʒ 

Is this what could have been considered the G6 or something?

No. The Power6 is the next generation of the Power5. The G5 was a scaled down version of the Power5, so if there is ever a "G6," it would be a scaled down version of the Power6.

Oh, and the PowerX chips are NOT PowerPC chips.

[Eug]

Originally Posted by Dakarʒ 

Is this what could have been considered the G6 or something?

It's possible a G6 or G7 could have been based off POWER6, but Macs likely wouldn't have used POWER6 itself (even if Macs were still on PowerPC).

For example, the G5 was based off POWER4, but no Macs used POWER4 itself. Mind you, one reason Macs couldn't use POWER4 cuz it had no Altivec. I believe POWER6 does actually support Altivec. However, another reason is cost.

Originally Posted by mdc 

I wonder with Universal Binaries if Apple would ever make a Mac Pro, or iMac with another PowerPC chip in it? Or are they Intel all the way now.

100% x86. I wouldn't say 100% Intel, since I could see Apple using AMD chips if Intel floundered.

[Dakarʒ]

Ah, thanks.

Edit: Thank you, both.

[Eug]

Originally Posted by Person Man 

No. The Power6 is the next generation of the Power5. The G5 was a scaled down version of the Power5, so if there is ever a "G6," it would be a scaled down version of the Power6.

The G5 was based off POWER4, not POWER5.

Oh, and the PowerX chips are NOT PowerPC chips.

They're all part of the "Power Architecture".

[Cipher13]

Originally Posted by Person Man 

The G5 was a scaled down version of the Power5, so if there is ever a "G6," it would be a scaled down version of the Power6.

Oh, and the PowerX chips are NOT PowerPC chips.

The G5 was based on the Power4 core, not the Power5 core.

Additionally, the PowerX (Power4, 5 and 6 at least) *are* PowerPC chips, in essence; they implement the PowerPC instruction set. What more do you want? Sure, you can argue they're not PowerPC because they weren't developed under AIM, but that's not very conducive to the conversation at hand.

[badidea]

More Power! Arrh! Arrh! Arrh!

[Chuckit]

Originally Posted by mdc 

I wonder with Universal Binaries if Apple would ever make a Mac Pro, or iMac with another PowerPC chip in it? Or are they Intel all the way now.

Apple certainly could. But there aren't really any PowerPC offerings that would tempt them to do so, OP notwithstanding.

[Person Man]

Originally Posted by Eug 

The G5 was based off POWER4, not POWER5.

I stand corrected.

They're all part of the "Power Architecture".

Yes, but they're still not the same chip.

[goMac]

Universal Binaries are a temporary solution for migrating over to Intel, not a permanent one. We're already starting to see Intel only software.

[mdc]

I hope they aren't a temporary solution since I don't see G4's and G5's vanishing any time soon. My PowerBook 1.67ghz G4 is still going strong and I could see myself with it for 2 or 3 more years.

[goMac]

Originally Posted by mdc 

I hope they aren't a temporary solution since I don't see G4's and G5's vanishing any time soon. My PowerBook 1.67ghz G4 is still going strong and I could see myself with it for 2 or 3 more years.

They're just like the FAT binaries back in the PowerPC days. While FAT binaries always stuck around, it wasn't too long until PowerPC only software was being released. Creative Software and games will probably be the first to go Intel only. Quite a few games are already Intel only, and portions of CS3 are Intel only. I would be willing to bet all of CS4 will be Intel only.

[Chuckit]

Originally Posted by goMac 

Universal Binaries are a temporary solution for migrating over to Intel, not a permanent one. We're already starting to see Intel only software.

Authoritative source or GTFO. The fact that some people use X11 doesn't mean Aqua is a temporary solution.

[Big Mac]

Originally Posted by Cipher13 

Additionally, the PowerX (Power4, 5 and 6 at least) *are* PowerPC chips, in essence; they implement the PowerPC instruction set. What more do you want? Sure, you can argue they're not PowerPC because they weren't developed under AIM, but that's not very conducive to the conversation at hand.

I hate to be technical, but your definition requires work. POWER chips are similar to PowerPC chips, but they aren't essentially the same. And that's not because of some artificial distinction that PowerPC chips were developed by a certain team like AIM - that doesn't even make sense because G5 development had nothing to do with Motorola. POWER chips preceded PowerPC chips. They are mainframe chips that implement the POWER architecture, which, IIRC, shares much in common with the PowerPC architecture but is not equivalent. PowerPC is the desktop derivative.

[goMac]

Originally Posted by Chuckit 

Authoritative source or GTFO. The fact that some people use X11 doesn't mean Aqua is a temporary solution.

No, but X11 should ideally be a temporary solution. Just like Universal Binaries, X11 provides a temporary solution for getting Linux and UNIX software working OS X. It's a transition environment. That said, there obviously is some X11 software that will never make it over to Aqua land. But for a lot of software, specifically OpenOffice, X11 is not the ideal environment and the developers are working towards an Aqua native version.

Support for the ability to use Universal Binaries will probably be around until the end of OS X. But I don't see many developers maintaining the PowerPC branches of their applications past the next year or two.

Apple will eventually drop PowerPC support in Mac OS X all together, just like 8.5 dropped support for the 68k chips four years after the first PowerPC came out. This would mean the end of Universal builds being the default option in XCode.

PowerPC support in a lot of cases is just too much work for most developers to do. I mean, it's not a problem for Joe Six Pack writing shareware apps, but for the big boys like Adobe, Microsoft, and Apple, supporting the PowerPC is going to be a pain. And they're going to be more eager to just dump PowerPC support the more market share that Mac OS X Intel gains.

I've asked some people at Apple's opinions on Universal binaries, but that's really not in the scope of this discussion. But I don't think Universal binaries are really on anyones mind as a permanent solution. Certainly they'll still be produced for a year or two more. But Universal binaries are not a forever and ever thing, as for bigger companies they take a lot more work to build.

[Big Mac]

Originally Posted by Chuckit 

Authoritative source or GTFO. The fact that some people use X11 doesn't mean Aqua is a temporary solution.

Unfortunately, he is more right than wrong. Surely Universal Binaries will be around longer than FAT binaries, but there will come a point when more and more software starts going Intel only. It's bound to happen given the fact that Apple went Intel only. Eventually Macs will be old enough and replaced in sufficient quantities with Mactels that companies (including Apple) will start to justify shipping Intel-only binaries. Just look back at the comments of that Adobe development team working on that Intel only OS X app they released where they said that Apple was telling them the (real) Mac market wasn't worth their development effort.

It would be awesome if IBM could make major strides and release a G6 that blows the then current Intel offerings out of the water. It would be poetic justice if Apple were forced to return home.

[TETENAL]

Originally Posted by Big Mac 

It would be awesome if IBM could make major strides and release a G6 that blows the then current Intel offerings out of the water. It would be poetic justice if Apple were forced to return home.

Apple would never return "home". The capability to natively run Windows has been shown to be a significant selling point for Macs. It would be stupid to give that up again. That awesome G6 would just sit in IBM's shelves and collect dust.

[Chuckit]

So basically, there's no good reason to believe Universal Binaries are going away any sooner than hardware of that generation would normally be phased out. Shiny.

[olePigeon]

It doesn't matter how fast the Power6 is, you're still not going to stick it into a laptop. That was Apple's problem and that's why they've moved to Intel.

[Big Mac]

Originally Posted by TETENAL 

Apple would never return "home". The capability to natively run Windows has been shown to be a significant selling point for Macs. It would be stupid to give that up again. That awesome G6 would just sit in IBM's shelves and collect dust.

Or maybe Apple could be courageous enough to support both architectures so that Mac users who have no interest in running Windows could buy real Macs again.

[goMac]

Originally Posted by Chuckit 

So basically, there's no good reason to believe Universal Binaries are going away any sooner than hardware of that generation would normally be phased out. Shiny.

Right, but they are going away. They aren't sticking around forever, and they aren't there to leave the PowerPC open as a option in the future.

[goMac]

Originally Posted by Big Mac 

Or maybe Apple could be courageous enough to support both architectures so that Mac users who have no interest in running Windows could buy real Macs again.

The PowerPC is not faster than the Intel chips at this point, and it would be 2-4 years until we'd see a Power6 derivative in a Mac. Apple wouldn't want to waste the time when Intel will have chips just as fast ready when a Power6 desktop derivative would be ready.

[Chuckit]

Originally Posted by Big Mac 

Or maybe Apple could be courageous enough to support both architectures so that Mac users who have no interest in running Windows could buy real Macs again.

What is more real about PowerPCs than Core processors?

[Cipher13]

Originally Posted by Big Mac 

I hate to be technical, but your definition requires work. POWER chips are similar to PowerPC chips, but they aren't essentially the same. And that's not because of some artificial distinction that PowerPC chips were developed by a certain team like AIM - that doesn't even make sense because G5 development had nothing to do with Motorola. POWER chips preceded PowerPC chips. They are mainframe chips that implement the POWER architecture, which, IIRC, shares much in common with the PowerPC architecture but is not equivalent. PowerPC is the desktop derivative.

So don't be unnecessarily technical. I was quite clear as to the bounds of my definition:

"...in essence; they implement the PowerPC instruction set."

That's no artificial distinction.

[Laminar]

Originally Posted by Big Mac 

Or maybe Apple could be courageous enough to support both architectures so that Mac users who have no interest in running Windows could buy real Macs again.

I can barely stand how crippled my not-real-Mac is.
I mean, I can't even do anything, like run OS X, iLife, etc etc.

[Dakarʒ]

I think the best part of this PowerPC vs Intel arguments is this quote

Apple and Intel: As kosher as a cheeseburger.

Yeah, what percentage of people care about kosher? It really is the perfect metaphor.

[Person Man]

Originally Posted by Big Mac 

I hate to be technical, but your definition requires work. POWER chips are similar to PowerPC chips, but they aren't essentially the same. And that's not because of some artificial distinction that PowerPC chips were developed by a certain team like AIM - that doesn't even make sense because G5 development had nothing to do with Motorola. POWER chips preceded PowerPC chips. They are mainframe chips that implement the POWER architecture, which, IIRC, shares much in common with the PowerPC architecture but is not equivalent. PowerPC is the desktop derivative.

Thanks for backing me up here.

Originally Posted by Big Mac

Or maybe Apple could be courageous enough to support both architectures so that Mac users who have no interest in running Windows could buy real Macs again.

*sigh*, not this again. The Intel-based iMac sitting on my desk at work is EVERY BIT AS MUCH A REAL MAC as my Power Mac G5 at home is. And the MacBook Pro I plan on buying in a few months will also be as real a Mac as my Power Macintosh 7600 as well. End of story.

[Eug]

Yeah, my iMac Core 2 Duo blows my friend's dual G5 Power Mac out of the water in terms of speed, and the iMac is dead silent.

And yeah, I agree that universal binaries are a temporary solution. I could fully see OS X 10.7 being Intel only, along with any Apple application from that era (2011?).

[Laminar]

I wouldn't be surprised if 10.6 was Intel only. Apple has pledged PowerPC support until, what 2010? If Leopard has a 3 year life span that puts 10.6 right at the Intel-only point.

[mduell]

300GBps is only from cache to execution units, all inside the core, right? At 5Ghz that's a 512-bit wide cache memory bus, which is twice as wide as Core 2 Duo.

[Big Mac]

Originally Posted by Dakarʒ 

I think the best part of this PowerPC vs Intel arguments is this quote. . . .

Yeah, what percentage of people care about kosher? It really is the perfect metaphor.

It's supposed to be slightly humorous. I am using the term kosher in the popular figurative sense, as in "genuine, legitimate" (the third definition in Apple Dictionary).

[Big Mac]

Originally Posted by Laminar 

I wouldn't be surprised if 10.6 was Intel only. Apple has pledged PowerPC support until, what 2010? If Leopard has a 3 year life span that puts 10.6 right at the Intel-only point.

When did Apple ever make such a pledge?

[C.A.T.S. CEO]

Originally Posted by Big Mac 

When did Apple ever make such a pledge?

At WWDC 2006 or MWSF 2006, I think

[goMac]

Originally Posted by C.A.T.S. CEO 

At WWDC 2006 or MWSF 2006, I think

I think it was WWDC 2005. I don't recall it being a pledge to support the PowerPC until 2010. I believe it was a goal to have completely migrated everything to Intel only by 2010. The entire keynote is fuzzy in my mind though. I had gotten up pretty early that morning to get breakfast and get in line.

[TETENAL]

Originally Posted by Big Mac 

I am using the term kosher in the popular figurative sense, as in "genuine, legitimate" (the third definition in Apple Dictionary).

Macs are as "legitimate" after the switch from PowerPC to Intel as they were after the switch from 68k to PowerPC.

[Chuckit]

Or the switch from Motorola's G4 to IBM's G5.

[Big Mac]

Originally Posted by TETENAL 

Macs are as "legitimate" after the switch from PowerPC to Intel as they were after the switch from 68k to PowerPC.

Wrong.

Originally Posted by Chuckit 

Or the switch from Motorola's G4 to IBM's G5.

False. But I'll let you guys labor under that delusion.

[CharlesS]

Originally Posted by Big Mac 

Wrong.

How about you give some, you know, reason as to why the switch to x86 is any different from the switch to PPC?

[Big Mac]

Originally Posted by CharlesS 

How about you give some, you know, reason as to why the switch to x86 is any different from the switch to PPC?

It's a very different beast. I have given a number of reasons why in previous threads, but perhaps they're lost on the average user. One reason why it's different is because with the PPC Apple still dared to follow the "think different" line on the hardware level. Apple dared to reach for superiority. With the defection to Intel, Apple gave up and sold out. And now people are attracted to Apple hardware because of its ability to run Windows - because they're buying glorified PCs.

Look, I'm an Apple purist, a zealot, and the last of a dying breed of those who believe in the Mac Way and a higher ideal for the platform than the corporate bottom line.

[olePigeon]

Originally Posted by CharlesS 

How about you give some, you know, reason as to why the switch to x86 is any different from the switch to PPC?

The 680x0 and PowerPC are both RISC processors. If I recall correctly, the PowerPC even had an embedded 680x0 emulator on the CPU itself. This allowed you to run native 680x0 binaries with little to no slowdown, and without recompiling anything for any specific architecture. Later, Apple introduced FAT binaries which allowed you to compile applications on a PowerPC based Macintosh and still run it on older 680x0 machines.

Moving from a RISC PowerPC architecture to a CISC x86 architecture is very different. There is no efficient or fast way to emulate PowerPC on a an x86 or vise versa.

Apple did do some fancy tricks with XCode to allow you to cross-compile with little work involved, but it essentially just makes two binaries; one for each architecture.

[TETENAL]

Originally Posted by Big Mac 

Apple dared to reach for superiority.

Apple's Intel-Macs are superior to their PowerPC-predecessors in practically any aspect.

Originally Posted by Big Mac 

One reason why it's different is because with the PPC Apple still dared to follow the "think different" line on the hardware level.

Were early PowerMacs that were still based on CHRP not "real Macs"? By your definition not.

[TETENAL]

Originally Posted by olePigeon 

The 680x0 and PowerPC are both RISC processors.

The 68k were CISC processors.

If I recall correctly, the PowerPC even had an embedded 680x0 emulator on the CPU itself.

No PowerPC ever had something like that. The system had a software 68k-emulator similar to Rosetta now.

Apple did do some fancy tricks with XCode to allow you to cross-compile with little work involved, but it essentially just makes two binaries; one for each architecture.

And that's the exactly how fat-binaries worked on 68k and PowerMacs.

So, technically, where is the difference that makes the previous transition "real" and this one not?

[Big Mac]

Originally Posted by TETENAL 

Apple's Intel-Macs are superior to their PowerPC-predecessors in practically any aspect.

Not in my opinion, but that wasn't my point either. My point was that Apple dared to be outside of the mainstream and compete with the near monopoly Intel has always enjoyed. Apple threw in the towel.

Were early PowerMacs that were still based on CHRP not "real Macs"? By your definition not.

The early Power Macs weren't based on CHRP. I think you've got your acronyms a bit mixed up. The first common hardware platform spec for the PPC was PREP, and Apple rejected it because it was not Mac-like enough. After that Apple promoted CHRP/PPCP, but it was not supported by other manufacturers.

[Eriamjh]

If Apple released a Mac with the Power6 chip in it, it would come with 512MB of PC133 RAM and a 120GB HD.

We heard all about how great the G3 was supposed to be. Then the G4. Then the G5. Ignore this news. No computer we would ever get to use will ever use it.

[Big Mac]

Originally Posted by Eriamjh 

If Apple released a Mac with the Power6 chip in it, it would come with 512MB of PC133 RAM and a 120GB HD.

That's a ridiculous comment. The last couple Mac revisions were all shipping with modern and competitive RAM and hard drive specs. Apple probably even held back with even better revisions to make its PCs look better.

We heard all about how great the G3 was supposed to be. Then the G4. Then the G5. Ignore this news. No computer we would ever get to use will ever use it.

The G3, G4 and G5 were all great chips. The G4 didn't scale well because of its design, and development of the G5 wasn't kept up after introduction at the pace Intel was moving. The PPC 970 was the desktop variant of the Power4. If there were demand for it, IBM would create a desktop variant of the Power6. Unfortunately, Apple sold out.

[CharlesS]

Originally Posted by Big Mac 

It's a very different beast. I have given a number of reasons why in previous threads, but perhaps they're lost on the average user.

Open by being condescending. Nice.

One reason why it's different is because with the PPC Apple still dared to follow the "think different" line on the hardware level. Apple dared to reach for superiority. With the defection to Intel, Apple gave up and sold out. And now people are attracted to Apple hardware because of its ability to run Windows - because they're buying glorified PCs.

When Apple switched to PPC, they did so because at the time, it actually was a superior technology with a lot more promise than x86. Unfortunately, x86 caught up. In both switches, the processor they were previously using was stagnating and harming the platform, and the switch was done for technological, not political, reasons.

Apple wasn't "thinking different" when they switched to PPC - in fact, they were banking on CHRP catching on and eventually being the same hardware that would run computers made by many different manufacturers - including PCs running the PPC version of Windows NT. Yes, Big Mac, one of the big features of the PowerPC Macs was going to be the ability to run Windows.

Look, I'm an Apple purist, a zealot, and the last of a dying breed of those who believe in the Mac Way and a higher ideal for the platform than the corporate bottom line.

Close by using the thalo.net people's catchphrase. Nice.

[CharlesS]

Originally Posted by olePigeon 

The 680x0 and PowerPC are both RISC processors.

Wrong. 680x0 was CISC.

If I recall correctly, the PowerPC even had an embedded 680x0 emulator on the CPU itself. This allowed you to run native 680x0 binaries with little to no slowdown, and without recompiling anything for any specific architecture.

You recall incorrectly. The emulator was software-based. In order to take advantage of the speed of the PowerPC, apps had to be recompiled. Otherwise, they'd run more slowly than on the older 680x0 processors - just like PPC apps on Intel. Unfortunately, almost all Mac apps were 680x0-based at the time the first Power Macs came out, including most of the OS, which caused the Power Macs to run embarrassingly slowly at launch time.

Later, Apple introduced FAT binaries which allowed you to compile applications on a PowerPC based Macintosh and still run it on older 680x0 machines.

FAT binaries were needed from the beginning, unless you didn't want your app to be able to run on older hardware.

Moving from a RISC PowerPC architecture to a CISC x86 architecture is very different. There is no efficient or fast way to emulate PowerPC on a an x86 or vise versa.

If I understand correctly, the core of the x86 processors has been basically RISC for some time, with a CISC compatibility layer over it. There's basically no difference anymore between "RISC" and "CISC" processors.

The only reason that the 680x0 emulator was as fast as it was was because the PowerPC was running at twice the clock rate of the older processors. The performance difference between Intel and PPC isn't as drastic (unless you're talking laptops), so Rosetta will be slower than a PPC Mac... for now. As the hardware gets faster and faster, not so much.

Apple did do some fancy tricks with XCode to allow you to cross-compile with little work involved, but it essentially just makes two binaries; one for each architecture.

FAT binaries essentially contained two binaries as well - only the PPC code was in the data fork, while the 680x0 code was in 'CODE' resources in the resource fork.

[Big Mac]

Originally Posted by CharlesS 

When Apple switched to PPC, they did so because at the time, it actually was a superior technology with a lot more promise than x86. Unfortunately, x86 caught up. In both switches, the processor they were previously using was stagnating and harming the platform, and the switch was done for technological, not political, reasons.

It was done so that Apple wouldn't have to be different anymore. So that it wouldn't have to compete anymore on the microprocessor front. So that it could offer its customers the same type of PC they could buy from Dell or HP, with some proprietary tweaks to prevent the OS from running on other PCs, so that Apple could have its cake and eat it to.

Apple wasn't "thinking different" when they switched to PPC - in fact, they were banking on CHRP catching on and eventually being the same hardware that would run computers made by many different manufacturers - including PCs running the PPC version of Windows NT.

Then Apple should have had CHRP ready from the outset and supported it. Instead, Apple rejected PREP and proceeded to create and destroy its own clone market.

Close by using the thalo.net people's catchphrase. Nice.

I never went to thalo; I only heard about it here. Those words are my own.