Comments on: The brains behind Apple's Rosetta: Transitive

Company supplies the engine for Rosetta, a crucial bridge for enabling the move to Intel-based computers.
Is Intel a safe bet for Apple security?

[Wolven Spectre]

This is going to hurt Apple in ways that Steve jobs never thought possible

There are so many things wrong with this deal, there has to be a catch... Steve Jobs isn't a stupid man andheusually lears from his mistakes and then profits by them. What, is he worried about Kyoto, is he so ticked off that IBM was playing favourites with Microsoft and the XBox360's cell processors that he will throw the baby out with the bathwater? If you think that they are listing all the sercuiry holes with Intel's new processors and chipsets here, THINK AGAIN. Oh, and Steve, if you are reading this... THINK DIFFERENT!

I have more at my blog where I did a special report on the keynote and the aftermath:

The Useless Information File. No information is Useless. http://wwwUIFile.BraveLog.com

Misinformation

The IBM Cell Processor belongs to the PS3, and has since it's first dicsussion back in 2001 between Sony and IBM.

And yes, Steve is probably a bit ticked that IBM wouldn't provide A: A 3GHz or faster chip for the Powermac line, let alone a G5 chip for a laptop. He's thinking bigger and broader and wants faster chips for all lines of the hardware his company builds.

I was freaking out at first, but after some calming, I have found that this will be a smart move for the future, and will probably yield price drops across the board. IF these chips make the new machiens and software run faster with the same stability of the PowerPC Chip, it's definately worth it. And besides, the PowerPC software will still be around for a couple of years.

As soon as a dual core or 2.5 GHz Powerbook is introduced witht eh Intel chip, I will have one.

They haven't tested this for the last 5 OS versions for nothing. I'm sur e they've done they're research and your "blog" tyupe journalism means nothing to them, or anyone who realizes what's going on.

[sanenazok]

watch out

the unsecure chipset is going to get you. FUUUUUUDDDDDD

[thomcarl]

Security Holes in chip sets?

The security holes that plague x86 belong to Microsoft, not Intel.

[open-mind]

How will it hurt?

Will much faster laptops hurt Apple?

Will Windows compatibility hurt Apple?

Will free "Intel Inside" marketing hurt Apple?

I don't think so.

[Wolven Spectre]

This is going to hurt Apple in ways that Steve jobs never thought possible

There are so many things wrong with this deal, there has to be a catch... Steve Jobs isn't a stupid man andheusually lears from his mistakes and then profits by them. What, is he worried about Kyoto, is he so ticked off that IBM was playing favourites with Microsoft and the XBox360's cell processors that he will throw the baby out with the bathwater? If you think that they are listing all the sercuiry holes with Intel's new processors and chipsets here, THINK AGAIN. Oh, and Steve, if you are reading this... THINK DIFFERENT!

I have more at my blog where I did a special report on the keynote and the aftermath:

The Useless Information File. No information is Useless. http://wwwUIFile.BraveLog.com

Misinformation

The IBM Cell Processor belongs to the PS3, and has since it's first dicsussion back in 2001 between Sony and IBM.

And yes, Steve is probably a bit ticked that IBM wouldn't provide A: A 3GHz or faster chip for the Powermac line, let alone a G5 chip for a laptop. He's thinking bigger and broader and wants faster chips for all lines of the hardware his company builds.

I was freaking out at first, but after some calming, I have found that this will be a smart move for the future, and will probably yield price drops across the board. IF these chips make the new machiens and software run faster with the same stability of the PowerPC Chip, it's definately worth it. And besides, the PowerPC software will still be around for a couple of years.

As soon as a dual core or 2.5 GHz Powerbook is introduced witht eh Intel chip, I will have one.

They haven't tested this for the last 5 OS versions for nothing. I'm sur e they've done they're research and your "blog" tyupe journalism means nothing to them, or anyone who realizes what's going on.

[sanenazok]

watch out

the unsecure chipset is going to get you. FUUUUUUDDDDDD

[thomcarl]

Security Holes in chip sets?

The security holes that plague x86 belong to Microsoft, not Intel.

[open-mind]

How will it hurt?

Will much faster laptops hurt Apple?

Will Windows compatibility hurt Apple?

Will free "Intel Inside" marketing hurt Apple?

I don't think so.

[System Tyrant]

Smart Move.

I know the Apple fans think Steve has lost his mind and probably is going to sink Apple this time. But I think their wrong.

What is the most heavily used and support processors on the market? x86. I think they are trying to compete. IBM has not allowed them such competition with the latest PowerPC chips. I think it is also a software decision as well. If software companies don't have to rewrite their code to support a different processor then we might start seeing more Mac supported software.

If Apple puts as much thought into designing x86 PC as they did the PowerPC you might just start seeing more and more Apple computers everywhere. I know this idea sucks, but if you could buy a Apple computer and run Windows on it along side Mac OS then it may also give people the insentive to dump Windows for Mac OS.

Frankly, Linux has been falling short on the Desktop market. I like linux, but it isn't a Windows replacement yet. Maybe Mac OS can pick up the slack and give Windows some real competition.

Then again maybe we are stuck with Windows for ever and everything else is wishfull thinking.

[David Arbogast]

Agreed

Although I generally agree with your statements, I think it needs to be reiterated that this is a LONG TERM strategy for Apple. All the details have obviously not been released, but the fact that Apple is not giving 100% of its consideration to technical arguments shows quite clearly that they are interested in making a wise business move that will benefit the company in the long term. LONG TERM is not 2 or 3 years, and this emulation strategy is obviously just the first step in a very long process. Wisely, Jobs is placing technical arguments second to business decisions. The business is finally deciding what approach is beneficial in the long term, and it is now leveraging its engineers to make it happen. Too often, the engineers try to tell the company what the best decision is, and its usually a failing strategy. I tend to agree with this guy... http://www.Inaniloquent.com

[NWLB]

Half way there.

I agree in that having a Mac that runs Windows without the issues emulation brings does finally make buying such a system viable. It isn't that I'd buy only that system. But I would add it to the other computers my family uses. In that, there is a great deal of value in this move. To start people using Mac software is really the best way to facilitate the migration to that system. It helps build a larger base of software support. It helps break misconceptions about Macs. This is all in the great interest Apple must have.

I'm still NOT going to buy a Mac anytime soon. However I'm watching this develop with a more open mind than in the past.

NWLB
****
http://www.nwlb.net

[System Tyrant]

More thoughts

David is right about not having a lot of technical details and about the Long Term goal of Apple.

If Apple can build a x86 system like they build a PowerPC systems now I am going to be very excited. I still remember going to an apple store that was very quiet and asking if the computer was on because it was so quiet. I also think that Apple spends more time and money to find the best way to build a system.

What I really hope for in the short term is Mac OSX available for the x86 platform regardless of computer manufacturer. It's not that it would do me any good to buy it, but I still would just to play around with it.

[System Tyrant]

Smart Move.

I know the Apple fans think Steve has lost his mind and probably is going to sink Apple this time. But I think their wrong.

What is the most heavily used and support processors on the market? x86. I think they are trying to compete. IBM has not allowed them such competition with the latest PowerPC chips. I think it is also a software decision as well. If software companies don't have to rewrite their code to support a different processor then we might start seeing more Mac supported software.

If Apple puts as much thought into designing x86 PC as they did the PowerPC you might just start seeing more and more Apple computers everywhere. I know this idea sucks, but if you could buy a Apple computer and run Windows on it along side Mac OS then it may also give people the insentive to dump Windows for Mac OS.

Frankly, Linux has been falling short on the Desktop market. I like linux, but it isn't a Windows replacement yet. Maybe Mac OS can pick up the slack and give Windows some real competition.

Then again maybe we are stuck with Windows for ever and everything else is wishfull thinking.

[David Arbogast]

Agreed

Although I generally agree with your statements, I think it needs to be reiterated that this is a LONG TERM strategy for Apple. All the details have obviously not been released, but the fact that Apple is not giving 100% of its consideration to technical arguments shows quite clearly that they are interested in making a wise business move that will benefit the company in the long term. LONG TERM is not 2 or 3 years, and this emulation strategy is obviously just the first step in a very long process. Wisely, Jobs is placing technical arguments second to business decisions. The business is finally deciding what approach is beneficial in the long term, and it is now leveraging its engineers to make it happen. Too often, the engineers try to tell the company what the best decision is, and its usually a failing strategy. I tend to agree with this guy... http://www.Inaniloquent.com

[NWLB]

Half way there.

I agree in that having a Mac that runs Windows without the issues emulation brings does finally make buying such a system viable. It isn't that I'd buy only that system. But I would add it to the other computers my family uses. In that, there is a great deal of value in this move. To start people using Mac software is really the best way to facilitate the migration to that system. It helps build a larger base of software support. It helps break misconceptions about Macs. This is all in the great interest Apple must have.

I'm still NOT going to buy a Mac anytime soon. However I'm watching this develop with a more open mind than in the past.

NWLB
****
http://www.nwlb.net

[System Tyrant]

More thoughts

David is right about not having a lot of technical details and about the Long Term goal of Apple.

If Apple can build a x86 system like they build a PowerPC systems now I am going to be very excited. I still remember going to an apple store that was very quiet and asking if the computer was on because it was so quiet. I also think that Apple spends more time and money to find the best way to build a system.

What I really hope for in the short term is Mac OSX available for the x86 platform regardless of computer manufacturer. It's not that it would do me any good to buy it, but I still would just to play around with it.

[Andrew J Glina]

Information Request

How well did the 68000 -> PowerPC emulator go? I first used Macs when the change first happened but I only used native software so I have no idea. Anyone?

[David Arbogast]

not very critical

Emulators never fare well in the long term, but lets keep in mind that the emulation strategy is the first step in a very long-term strategy. The migration will not be easy, but it has to start somewhere.

[samkass]

68k->PowerPC

The transition from 68k to PowerPC went very smoothly for users. The emulator was about 40-60% of native speed, but so many key things were native (where an emulated app calls a native (fast) system library), and since an application spent most of its time in system libraries, that applications often ran closer to 80-90% of native speed. The actual speed of emulation is a bit of a red herring when you're running an application that, natively, spends most of its time in system libraries.

The painful, painful part of the 68k->PowerPC transition was for developers. Back in those days, direct memory acces was common, MacOS had no facility for multiple architectures, function pointers and memory references were all 68k-specific things, etc. With MacOS X, for large portions of it, this is a port BACK to x86, and all of those problems were solved long ago with Carbon and Cocoa.

[Earl Benser]

Classic was adequate

As noted by other responders, there were some techniques used
in pre-OS X days that wouldn't fly, but for he most part, pre-OS
X apps were well run, and with the faster processors, there was
not much noticeable speed loss.

Going back further, the transition from 68000 code to PPC code
again left some apps in the dust. Usually the software
companies came up with upgraded versions. Some apps were
just abandoned.

Full details have slipped my memory, but from my beginnning
on a Fat Mac, with multiple model changes following, I never felt
constrained by the processor changes, What effec tthere was was
dwarfed by the impact of software manufacturers' market driven
decisions aboutt product survival

[Andrew J Glina]

Information Request

How well did the 68000 -> PowerPC emulator go? I first used Macs when the change first happened but I only used native software so I have no idea. Anyone?

[David Arbogast]

not very critical

Emulators never fare well in the long term, but lets keep in mind that the emulation strategy is the first step in a very long-term strategy. The migration will not be easy, but it has to start somewhere.

[samkass]

68k->PowerPC

The transition from 68k to PowerPC went very smoothly for users. The emulator was about 40-60% of native speed, but so many key things were native (where an emulated app calls a native (fast) system library), and since an application spent most of its time in system libraries, that applications often ran closer to 80-90% of native speed. The actual speed of emulation is a bit of a red herring when you're running an application that, natively, spends most of its time in system libraries.

The painful, painful part of the 68k->PowerPC transition was for developers. Back in those days, direct memory acces was common, MacOS had no facility for multiple architectures, function pointers and memory references were all 68k-specific things, etc. With MacOS X, for large portions of it, this is a port BACK to x86, and all of those problems were solved long ago with Carbon and Cocoa.

[Earl Benser]

Classic was adequate

As noted by other responders, there were some techniques used
in pre-OS X days that wouldn't fly, but for he most part, pre-OS
X apps were well run, and with the faster processors, there was
not much noticeable speed loss.

Going back further, the transition from 68000 code to PPC code
again left some apps in the dust. Usually the software
companies came up with upgraded versions. Some apps were
just abandoned.

Full details have slipped my memory, but from my beginnning
on a Fat Mac, with multiple model changes following, I never felt
constrained by the processor changes, What effec tthere was was
dwarfed by the impact of software manufacturers' market driven
decisions aboutt product survival

[RetiredMidn]

This topic is poorly covered

The story is vague as to what the Rosetta technology is doing, and throws around references to translation and emulation as thought they are the same thing. They're not. And it's not clear to me which category Rosetta fits into.

Emulation is execution of "foreign" binaries in place; this is what Apple 68K emulator has done since the introduction of PowerPC in 1994. There is a performance hit (although modern PowerPC processors can now run 68K binaries faster than they ever ran on a 68K Mac).

Translation is conversion of foreign binaries to "native" binaries. Back during the 68K-to-PowerPC transition, Apple worked with a company (Echo Logic) to offer binary translation capabilities to ISVs. The technology wasn't widely successful, but the reasons had at least as much to do with changes in APIs and runtime architectures as it did with the instruction sets; the PowerPC-to-Pentium conversion is a very different problem. But what's interesting to note is that some of the binaries produced by the Echo Logic translation were actually more efficient than the original code.

I could go on at lenth, but I'm not getting paid to do it... ;-)

[RetiredMidn]

This topic is poorly covered

The story is vague as to what the Rosetta technology is doing, and throws around references to translation and emulation as thought they are the same thing. They're not. And it's not clear to me which category Rosetta fits into.

Emulation is execution of "foreign" binaries in place; this is what Apple 68K emulator has done since the introduction of PowerPC in 1994. There is a performance hit (although modern PowerPC processors can now run 68K binaries faster than they ever ran on a 68K Mac).

Translation is conversion of foreign binaries to "native" binaries. Back during the 68K-to-PowerPC transition, Apple worked with a company (Echo Logic) to offer binary translation capabilities to ISVs. The technology wasn't widely successful, but the reasons had at least as much to do with changes in APIs and runtime architectures as it did with the instruction sets; the PowerPC-to-Pentium conversion is a very different problem. But what's interesting to note is that some of the binaries produced by the Echo Logic translation were actually more efficient than the original code.

I could go on at lenth, but I'm not getting paid to do it... ;-)

[grossph]

Remember apple produces their own mother boards

We keep hearing about emulation/translation in software. Did anyone ever think, that maybe apple will produce a chip to be placed on their mother board that would to the translation. This would increase the speed of running NON native code by a factor of 2 or 3. Heck, maybe that was part of apples deal with intel "Hey intel, you want us to use your chip, then we need you to design a translation chip for us"

I assume Steve Jobs is a pretty smart guy and that there are a lot of smart guys at apple (though if they could use an extra one, I am available for hire :-) I am pretty sure they have thought about all these things.

No one really knows the what the deal is with Intel. It is going to take 2 years to make the transistion, so I am sure there is going to be some "updates" to the mother boards. Apple is only going to be using the Intel CPU, they still design all there other chipsets. Not that much different then their relationship with IBM. IBM made the CPU, apple design the mother board, etc.

My two cents

Philip Grossman
Senior Information Architect

[Earl Benser]

A Rosetta support chip isn't a good idea

A hardware solution is permanent non rewritable complication.
Inthe video world, software decompressors are much better that
hardware versions. They are comparable in speed, and can be
changed instantly. Rosetta is in the same boat. Once installed, it's a
sure bet that newer versions will occur. A chip solution leaves you
stuck. A software solution leaves you running up to date.

[grossph]

Remember apple produces their own mother boards

We keep hearing about emulation/translation in software. Did anyone ever think, that maybe apple will produce a chip to be placed on their mother board that would to the translation. This would increase the speed of running NON native code by a factor of 2 or 3. Heck, maybe that was part of apples deal with intel "Hey intel, you want us to use your chip, then we need you to design a translation chip for us"

I assume Steve Jobs is a pretty smart guy and that there are a lot of smart guys at apple (though if they could use an extra one, I am available for hire :-) I am pretty sure they have thought about all these things.

No one really knows the what the deal is with Intel. It is going to take 2 years to make the transistion, so I am sure there is going to be some "updates" to the mother boards. Apple is only going to be using the Intel CPU, they still design all there other chipsets. Not that much different then their relationship with IBM. IBM made the CPU, apple design the mother board, etc.

My two cents

Philip Grossman
Senior Information Architect

[Earl Benser]

A Rosetta support chip isn't a good idea

A hardware solution is permanent non rewritable complication.
Inthe video world, software decompressors are much better that
hardware versions. They are comparable in speed, and can be
changed instantly. Rosetta is in the same boat. Once installed, it's a
sure bet that newer versions will occur. A chip solution leaves you
stuck. A software solution leaves you running up to date.

cnet should see the advice of more compentent analysts and journalists

Illuminata analyst Gordon Haff wrote and i quote, "History says that binary translation basically doesn't work."

This guy is either ignorant or a moron. Not only has binary translation been done before successfully, but it was Apple that did it!

Now on to the reason to hire new journalists. Shortly there after in the same article you fine and I quote, "Apple, though not known for bending over backward to support users of older Macs, has some experience helping users with processor transitions. When it changed from Motorola 680x0 processors to PowerPC in 1994, it included emulation software that would let users run the older software on the newer machines."

This emulation software the author refers to is none other than binary translation software. Here the author of the article contradicts his own analyst but never bothers to point it out. One could wonder if the author of the article even bothered to read what the analyst wrote! Talk about shoddy journalism, you guys need a new journalist, a new editor, or both!

[Byronic]

I completely concur

This is ridiculous. Apple is *well* known for the 68k to PPC
transition. It was REMARKABLY successful. By the second
round of PPC machines (mine was a lowly 8100) all the 68k stuff
was running faster than it ever did natively.

Also, all the stories about how Apple looses customers with
every transition are totally inacurrate. The losses are much
more closely correlated with M$ and Intel successes. Intel have
made a great deal more out of the old x86 than we ever thought
possible (and did a lot better than it looked like they ever did in
the mid nineties when PPC was completely eclipsing Pentiums.
Of late, PPC has been doing well to keep up, and our hardware
advantages (chip advantages at least) have dwindled.

But the 68k emulation??? It all STILL RUNS TODAY!!! EVEN IN
CLASSIC!!! Bzzzzzt! Buy a clue.

I bet there are some emergency meetings at M$, and I bet the
Mac BU are not the ONLY people there that are going to be
looking at their new $999 development units. I hope Apple
sends that to them welded shut with a BOMB inside set to go off
if they pry it open. And, to be fair, a warning label. Intel
inside... :-)

[Thomas, David]

Right On

It is odd to find few, and I mean a few as in maybe three, CNet
writers who truly understand what they write about.

All of the others are simply hacks trying to make a buck.

cnet should see the advice of more compentent analysts and journalists

Illuminata analyst Gordon Haff wrote and i quote, "History says that binary translation basically doesn't work."

This guy is either ignorant or a moron. Not only has binary translation been done before successfully, but it was Apple that did it!

Now on to the reason to hire new journalists. Shortly there after in the same article you fine and I quote, "Apple, though not known for bending over backward to support users of older Macs, has some experience helping users with processor transitions. When it changed from Motorola 680x0 processors to PowerPC in 1994, it included emulation software that would let users run the older software on the newer machines."

This emulation software the author refers to is none other than binary translation software. Here the author of the article contradicts his own analyst but never bothers to point it out. One could wonder if the author of the article even bothered to read what the analyst wrote! Talk about shoddy journalism, you guys need a new journalist, a new editor, or both!

[Byronic]

I completely concur

This is ridiculous. Apple is *well* known for the 68k to PPC
transition. It was REMARKABLY successful. By the second
round of PPC machines (mine was a lowly 8100) all the 68k stuff
was running faster than it ever did natively.

Also, all the stories about how Apple looses customers with
every transition are totally inacurrate. The losses are much
more closely correlated with M$ and Intel successes. Intel have
made a great deal more out of the old x86 than we ever thought
possible (and did a lot better than it looked like they ever did in
the mid nineties when PPC was completely eclipsing Pentiums.
Of late, PPC has been doing well to keep up, and our hardware
advantages (chip advantages at least) have dwindled.

But the 68k emulation??? It all STILL RUNS TODAY!!! EVEN IN
CLASSIC!!! Bzzzzzt! Buy a clue.

I bet there are some emergency meetings at M$, and I bet the
Mac BU are not the ONLY people there that are going to be
looking at their new $999 development units. I hope Apple
sends that to them welded shut with a BOMB inside set to go off
if they pry it open. And, to be fair, a warning label. Intel
inside... :-)

[Thomas, David]

Right On

It is odd to find few, and I mean a few as in maybe three, CNet
writers who truly understand what they write about.

All of the others are simply hacks trying to make a buck.

[M A]

Patching an Intel processor???

Umm..I'm starting to think that when Apple users think of Intel & "security" they think that all those patches being released are for Intel hardware. Microsoft software & their security has nothing to do with Intel hardware. Really, have you ever heard of patching your processor because of bugs?? Or chipset for that matter.

[Thomas, David]

True

For some reason, and its not mac people is the average computer
user, people think the chip set is related to todays security issues.
It isn't Who knows why, maybe some sort of techno-urban rumor.

Intel has had its missteps that they have learned from. Remember
the divide by zero error that gave you an answer of 256?!

Oh I wish I had a crystal ball.

[M A]

Patching an Intel processor???

Umm..I'm starting to think that when Apple users think of Intel & "security" they think that all those patches being released are for Intel hardware. Microsoft software & their security has nothing to do with Intel hardware. Really, have you ever heard of patching your processor because of bugs?? Or chipset for that matter.

[Thomas, David]

True

For some reason, and its not mac people is the average computer
user, people think the chip set is related to todays security issues.
It isn't Who knows why, maybe some sort of techno-urban rumor.

Intel has had its missteps that they have learned from. Remember
the divide by zero error that gave you an answer of 256?!

Oh I wish I had a crystal ball.

[Richard G.]

Minimal performance impact

It's true that emulation software slows down the performance of the original software. So there will be 3rd-party applications that run slower on the intel-based macs.

However, a lot of software spends it's time running inside system libraries or the kernel software itself. That software will be compiled native to the intel chip. So, there won't be quite as big a performance hit as what's feared.

Another thing: the major 3rd-party software vendors for apple (most notably Adobe) have already voiced their support to port and recompile their software for native intel chips. And apple will certainly port it's own applications to navtive intel. So most users will be running native software anyways, as long as they pay for the Adobe upgrades.

[Richard G.]

Minimal performance impact

It's true that emulation software slows down the performance of the original software. So there will be 3rd-party applications that run slower on the intel-based macs.

However, a lot of software spends it's time running inside system libraries or the kernel software itself. That software will be compiled native to the intel chip. So, there won't be quite as big a performance hit as what's feared.

Another thing: the major 3rd-party software vendors for apple (most notably Adobe) have already voiced their support to port and recompile their software for native intel chips. And apple will certainly port it's own applications to navtive intel. So most users will be running native software anyways, as long as they pay for the Adobe upgrades.

[originalbitman]

IntelaMac

Apple developers respond:

http://www.macworld.com/news/2005/06/08/saying/index.php

Thank you

That si a very re-assuring article to hear the developer's spekaing so positively about it.

[originalbitman]

IntelaMac

Apple developers respond:

http://www.macworld.com/news/2005/06/08/saying/index.php

Thank you

That si a very re-assuring article to hear the developer's spekaing so positively about it.

[Al Feldzamen]

"Smart Move" - "Smart Writer???"

"I think their wrong." ??? Send this guy back to high school (or
grammar school) English class !!

[Andrew J Glina]

Give it a rest

This is not grammer school. Try to listen to peoples opinions instead of picking on their English skills.

[Al Feldzamen]

"Smart Move" - "Smart Writer???"

"I think their wrong." ??? Send this guy back to high school (or
grammar school) English class !!

[Andrew J Glina]

Give it a rest

This is not grammer school. Try to listen to peoples opinions instead of picking on their English skills.