OS X x86 FAQ

[pawnstar]

snip
Are there any insecurities in the x86 chip that aren't in the PPC that will be exploited under Mac OS X?
-Chase

I don't know if it classes as insecurity, but in the universal binary PDF it says on page 22:

"…programming errors, or other operations that access past the end of a local variable array or otherwise incorrectly manipulate values on the stack may be more likely to crash applications on x86 systems than on PowerPC."

Overall it seems like PPC is better but what's the point if nobody develops the chips Apple (and we) need.

As to overall security, FreeBSD (OS X deep underneath) runs x86 as do other systems. The security is dictated by the OS, not the processor.

 

[efoto]

Things don't touch in parallel

Just saw the best question EVAR on another site:

With the changeover, does that mean the new ''Macintels'' will ship with parallel ports?

Please no one answer that....its not even worth the two letters needed to respond.

 

[dubbz]

Ewwy, not sure I like that idea.

I am hoping that Apple (given their five years working already) already have/know a way to make to generic PC unable to run OSX and therefore keep their OS at least mostly pirate free and keep it great. I think the more likely the OS is to be pirated the quicker development goes down the tubes (due to lower income cause no one is buying) and then we end up with swiss-cheese Windows, which who needs another holy OS?

Even if they're not using OpenFirmware, doesn't mean they'll use a standard Award or Phoenix or whatever BIOS. Intel has its own boot system (which I don't know the specifics of, nor do I remember the name...)

But if they they don't want to put blocks in the way of a dualboot system with Windows, they'll probably have to stick with a fairly standard BIOS (but I'm sure they'd tweak it somehow).


/Don't quote me on this, though. I'm just talking out of my behind

 

[efoto]

Even if they're not using OpenFirmware, doesn't mean they'll use a standard Award or Phoenix or whatever BIOS. Intel has its own boot system (which I don't know the specifics of, nor do I remember the name...)

But if they they don't want to put blocks in the way of a dualboot system with Windows, they'll probably have to stick with a fairly standard BIOS (but I'm sure they'd tweak it somehow).


/Don't quote me on this, though. I'm just talking out of my behind

I hope they tweak it in some manner or another to at least set it far apart from current boot BIOS systems as I find them ill-conceived and downright annoying. I'm not talking boot-post screens or anything, the actual navigation and use of a bios is just worthless is all cases I have seen.

 

[dubbz]

^^^

I whould think so. The current BIOSes is nice for people like me, who like to tweak, but stuff like memory timings, voltages settings and what not are just not required for most people. They could remove lot of the more advanced stuff without problems.

 

[bit density]

Will Macs switching to Intel processors mean viruses will be coming to Mac?

More likely. Buffer overflows can all use the same code on x86 machines, this is why they are all dangerous. The nature of taking advantage of a buffer overflow *is* chip specific and very well understood on the x86. There *have* been buffer overflow opportunities in OSx (in libraries that OSX has shipped with).

The default user on the mac HAS root privelidges. (IE you can do pretty do anything you want as a user, and any "virus" will).

The Mac system also suffers from a likely and very dangerous vector. Most Mac users use the same mail client and adx book. Mac users tend to know other mac users. And Mac users have a high degree of trust because there have not been any viruses to date.

There will be a virus on the mac.

Are my Power-based Macs all obsolete and useless now?
No not obsolete now. But will become so. Universal binaries will create Testing issues for companies. They *will* find platform specific bugs, and both of these things cost companies money.

However, the plan for moving less expensive machines first, will mean that companies will be more encouraged to not ship x86 only binaries. This will be *less* true when x86 top of the line machines ship. (These customers are the most likely to upgrade). Some point soon after that you are likely to see x86 only binaries...

What about my older PPC apps? What if they're not recompiled for new Intel Macs?

This MAY make a difference for high performance requirement apps that have been "legally" deprecated. Such as DVD copiers.

Is this going to be a huge, painful transition?

Likely not. It depends on how often you update your software. It may very well make a difference if you don't, yet you do update you machine on a 2-3 year cycle, and expect your machine and your software to run faster. For instance, it is NOT likely that PS, Indesign, or Office will run as fast on a first generation x86 machine as a 1-2 generation old PPC. It is likely that there will be updates that DO run as fast, but if you are not buying the update you may be disappointed.

But for most people, there will be little to no difference.



Will this really be a seamless transition?

Not likely, there WILL be issues. You will have to upgrade your software for instance to the universal binaries. And some of the software will not be ready when you are.

Some stuff will wierdly run slow.

Some paralegal stuff will probably stop working.

But most things will work. This will make the stuff that doesn't work more noticeable.

Will there be an "Intel Inside" sticker on future Macs?

5 bucks a machine may be too hard to turn down.

What does this mean for Apple's competition and other groups?

There are new things to consider with the use of an Intel platform, all of it speculation:

It is likely that OSx will run on wintel boxes. It already does, and the version that is going into developer hands does just that. Expect it on Torrents shortly thereafter.

Simply it will cost Apple a ton of money to try and keep OSx off of wintel boxes, and it is nearly gauranteed NOT to work. Expect more PAID customers who walk into an Apple store to see what they can get with their Ipod, and walk out with a copy of OSx. Increased Piracy issues are most likely non-customers. But putting OSx on any machine just makes Apple Money, in addition to the money they get from thier own hardware sales. As a matter of fact, the money they make will be about the same.

Microsoft will NOT leave X86. That is pure and utter poppycock. They are on PPCesque platform because they provide the hardware, and they own the chip design which they can move to any manufacturer, which *will* allow volume discounts that they did not get on Xbox I. But Windows and Longhorn are forever married to the X86 Platform.

 

[madget]

This is a very informative thread -- thanks to all!

Will the new machines with the intel chips be able to run Classic mode? (Yes, I have one remaining program which I use in Classic...)

 

[stcanard]

I'm going to give a hit at some of these answers as well. The first point to note is that Intel made a huge shift in their processor philosophy a couple of years ago (we're just seeing it now, but these things have to have been in the pipeline for a long time), that has actually nullified a lot of the differences between x86 and PowerPC architecture.

For a really detailed look at what is going on in intel, you can read these aricles:

A Look at Centrino's Core: The Pentium M
Multicore, dual-core, and the future of Intel

But be warned they are very technical. It will, however answer a lot of your questions about why.

First point to remember is x86's haven't been CISC for a long time. Both Intel and AMD have pipelines that break the instruction set down into micro (RISC-like) instructions that gains a lot of the advantages of a RISC chip.

I thought that the PPC was built for multi-processing more than x86, so multiple processors made more of a difference with Macs than PCs. Is it PPC that's better, Mac OS X, or both?

Realistically, no. Both have had built in support for a long time, and there probably isn't much to differentiate them, aside from the fact that Intel is already shipping dual-core (like) chips, whereas IBM and Freescale are promising them "real soon now", probably around about the time they hit 3.0GHz

This doesn't quite make sense to me. The PPC is known for using less power than its x86 counterparts, so where did Steve get that moving to x86 is going to require less power?

This used to be true, but everything changed with the introduction of the PPC-970 (aka G5), and the Pentium-M core (aka Centrino). The 970 has become a power hog requiring complex cooling (hence water cooled powermacs and no G5 PowerBook), whereas the Pentium-M core has become the standard for power efficiency, allowing Centrino based notebooks to kick a Mac notebook's proverbial a** when it comes to battery life.

Not only that, but Intel has a roadmap with these things getting faster, cooler, and dual core, whereas IBM has a promise to build a lot of chips for game consoles, and will look into the power issue whenever they get some spare time.

I thought that in the end, the PPC roadmap was better in that the PPC had more room to grow than x86. While not growing as rapidly, it would still surpass x86 when technical limitations were met. What is Intel showing Steve that is making him so confident that Intel is the way to go? Are we in for another OS switch 5 years from now, and then a chip switch in 10 years as technology changes?

Essentially, I think, Intel is showing Steve a roadmap that heads in a direction he wants, whereas IBM isn't showing much of a roadmap at all, and with their fab plants going in the direction of all 3 next-gen consoles, I'm sure Steve is looking forward to a time when IBM doesn't really care about them anymore.

As a side benefit, when was the last time you heard about Dell having to backorder systems because Intel couldn't provide enough chips? This whole idea of the constant announce, delay, delay, delay probably loses Apple more potential business than this change might!

Are there any insecurities in the x86 chip that aren't in the PPC that will be exploited under Mac OS X?

No. All security issues are OS level and above. Especially since x86 now provides a no-execute bit just like the PPC.

 

[mischief]

The default user on the mac HAS root privelidges. (IE you can do pretty do anything you want as a user, and any "virus" will).

Not entirely accurate. The default user has some Root priviledges by standard *nix definitions but it is more accurate to say that the main user has root priviledges only momentarilly when initiating an action (like an install) that requires authentication. This is a GUI equivalent of a "sudo" Terminal action. In the most critical instances Root access requires creation of a true Root account and authentication in that capacity.

Simply it will cost Apple a ton of money to try and keep OSx off of wintel boxes, and it is nearly gauranteed NOT to work..... But putting OSx on any machine just makes Apple Money, in addition to the money they get from their own hardware sales. As a matter of fact, the money they make will be about the same.
.

As pointed out above, there are enough features on Apple mobos such as bootability through the FW controller, the PMU, the custom ASIC that controls the FSB, etc. to keep the platform closed. Attempting emulation of one such feature could be done in a finite amount of time, Atempting emulation of a dozen or so small and large, gross and subtle features could be close enough to insurmountable to be effective.

This is amplified by the real potential for frequent revving of custom chipsets due to Intels relatively rapid turnaround and stupidly broad base of device-controller comatibilities. I find it very likely that Apple's Mobos will be a joint design built by Intel Fab facilities resulting in a HW designer's wet dream in the '06 series PM.

 

[slb]

And Mac people are shocked when they see these exact things working just fine on a Sony VIAO. It's not just an Apple thing. It's the choice of the manufacturer to do it well (e.g. IBM/Lenovo ThinkPads, or some features (but not quality) on Sonys) or to do it sloppily (e.g. most Dells and all Gateways).

My point was that every notebook manufacturer does it a different way, using different hardware calls, and different buttons. My HP laptop also had those extra buttons, but you needed to install special software for it to work. Whenever I did a clean install of Windows, the buttons were useless and did nothing.

OS X would have to support every possible configuration for everybody's extra laptop functionality to work, and since it's a hodge-podge of no standards, it's a mess.

 

[admanimal]

According to the PDF that's available from apple, Intel-based Macs will NOT use OpenFirmware. They are effectively standard Intel boards, including BIOS.

Whoa easy there killer. I think you're reading more than it says on the page. All it says is that they won't use Open Firmware. What they will use is not mentioned anywhere.

 

[bit density]

Whoa easy there killer. I think you're reading more than it says on the page. All it says is that they won't use Open Firmware. What they will use is not mentioned anywhere.

The 1.4.1 disks that are being handed out to developers run on standard wintel boxes.

 

[admanimal]

The 1.4.1 disks that are being handed out to developers run on standard wintel boxes.

Who/Where says that? I know I have heard that those P4 dev machines Apple is renting are actually modified PM hardware that has been made to work with Intel CPUs.

 

[mischief]

The 1.4.1 disks that are being handed out to developers run on standard wintel boxes.

The run-standards for the SDK's for an as-yet unpublished version of the OS on an as-yet unbuilt hardware platform are moot as regarding the final firmware config unless you care to dig into the compiler's settings.....

This is very much like similar arguements that were made about the OS9 to OS X migration path. Dig into the manuals, Backtrace the compiler's actions in regards to firmware presets but please do not suppose in this thread without verifiable evidence. We're going to have enough trouble with bad info without supposition to add to it.

 

[bit density]

As pointed out above, there are enough features on Apple mobos such as bootability through the FW controller, the PMU, the custom ASIC that controls the FSB, etc. to keep the platform closed. Attempting emulation of one such feature could be done in a finite amount of time, Atempting emulation of a dozen or so small and large, gross and subtle features could be close enough to insurmountable to be effective.

The 1.4.1 disks that will be given to developers will run on standard wintel boxes.

The emulation described above, is less emulation than API mapping. A trivial thing to hack. And you give little credit to the tenacity of folks that will hack this. Resulting in pirates only running on Wintel boxes. And this advantages Apple and their potential customers how?

Again, it is easier, costs less, makes more money to allow the os to run on a standard wintel box. Apple makes money on every legally purchased copy. They give it away on their own hardware, charge you if you want to run it on other hardware.

Apple competes on price advantage and quality of hardware. Until ofcourse they get to 90 percent market share and someone cries foul. But until then, Apple wins and Customers win. There is no reason to keep the OS xenophobically on your own hardware.

The only requirement will be no legacy hardware support. And if Sony decides to not write drivers for all their extra buttons, then all the extra buttons don't work. Simple enough.

 

[admanimal]

The 1.4.1 disks that will be given to developers will run on standard wintel boxes.

Please prove it.

 

[bit density]

This is very much like similar arguements that were made about the OS9 to OS X migration path. Dig into the manuals, Backtrace the compiler's actions in regards to firmware presets but please do not suppose in this thread without verifiable evidence. We're going to have enough trouble with bad info without supposition to add to it.

Why would you believe that Apple is going to lock the OS to their hardware? What is the advantage to Apple for doing so? You are using as much supposition as I am. I think my supposition of less cost, more money makes more sense than xenophobic, nearly paranoid attempts to keep the system locked up.

Why leave the wintel version to the pirates and the hackers. Why not let paying customers put it on a wintel box?

(Note, I am NOT saying that they should license the OS to Dell. I am saying a customer should be able to walk into an Apple store and buy the OS).

 

[admanimal]

Why would you believe that Apple is going to lock the OS to their hardware?

Because this is exactly what Phil Schiller said they were going to do.

There are many practical reasons why they would want to do this. One of the more altruistic ones is because if they can control the hardware OS X is installed on, they can provide better device driver support.

 

[Xtremehkr]

Just wondering what this new chip will do for (or, "to") the current iMac design? Will we still see desktop performance from a desktop iMac or are we going to see laptop performance from a laptop CPU housed in a desktop iMac?

Any thoughts on this?

I think in the next two years the iMac will continue to be updated. It will probably get the G5 2.3 next and may finish with the 2.7 before going to Intel chips.

I don't see why a desktop like the iMac would get a laptop processor though, there is no need for it to use less power. Given how much heat the G5s put out, the iMac could be even more compact with an Intel processor.

Given that the iMac is not the forerunner performance, it may stay with PPC longer than Powermacs do.

 

[mahashel]

x86 over Itanium for now?

I initially thought this announcement would mean Itanium PowerMacs, which sounded pretty nifty to me.
However, I suppose the porting problems from moving PPC code to IA-64 would be much more complex and damaging on the Mac software front than shifting to x86.
Too bad.. the online reports I've read about the latest IA-64 make it sound pretty beefy. Maybe we can kill all the G5 Powerbook threads and start an Itanium-Powerbook thread?!

 

[bit density]

Please prove it.

Oh, come on.

But lets play the game. The developers are signing up for an x86 1.4.1 *DISK*, not a machine, not a hundred machines, but DISKS. What the hell do you think they are going to run it on?

 

[admanimal]

Oh, come on.

But lets play the game. The developers are signing up for an x86 1.4.1 *DISK*, not a machine, not a hundred machines, but DISKS. What the hell do you think they are going to run it on?

Can you please point me to where there is proof that they are just giving it out on disks to developers, then? I know they can rent one of those dev machines with Tiger for Intel, but I have not heard anyone say they can just get a disk.

During the keynote, all Steve mentions that developers can get on CD is XCode 2.1.

 

[stcanard]

Oh, come on.

But lets play the game. The developers are signing up for an x86 1.4.1 *DISK*, not a machine, not a hundred machines, but DISKS. What the hell do you think they are going to run it on?

Uhh, huh?

According to the press release, for $999 they are getting a *disk* and a machine to run it on

Trolling's a lot more effective if you do your reading first.

 

[savar]

But that would render it almost useless... and defeat the purpose of having it in the first place.
Without this a lot of people won't immediately switch until all their software is ported.

Another question I have is:
Do fat binaries run slower than slim ones?

No, but they do take up more space on disk. I hope the installers will make a novice assumption and install only the one binary that the users *needs* on his particular computer.

 

[slb]

Will Macs switching to Intel processors mean viruses will be coming to Mac?

More likely. Buffer overflows can all use the same code on x86 machines, this is why they are all dangerous. The nature of taking advantage of a buffer overflow *is* chip specific and very well understood on the x86. There *have* been buffer overflow opportunities in OSx (in libraries that OSX has shipped with).

A fair statement, but compiler technologies and chip technologies are making buffer overflows a non-issue.

The default user on the mac HAS root privelidges. (IE you can do pretty do anything you want as a user, and any "virus" will).

The default user on OS X runs in a non-root account, and you can't do whatever you want. This is why you have to enter a password to install apps.

By default, the root account isn't even enabled on OS X.

Microsoft will NOT leave X86. That is pure and utter poppycock. They are on PPCesque platform because they provide the hardware, and they own the chip design which they can move to any manufacturer, which *will* allow volume discounts that they did not get on Xbox I. But Windows and Longhorn are forever married to the X86 Platform.

Microsoft already has a PPC version of Windows running on the X-Box 360. .NET provides processor independence akin to Java, making x86 reliance a non-issue. The X-Box 360 is using a PowerPC-based processor, and Microsoft has been very interested in the hardware market the past few years because they want to sell both hardware and software, and the X-Box 360 is their first step towards doing that. I think it's obvious.