Mac OS X Intel Developer Box Feedback

[goof_ball]

I think Apple should make sure these can run classic. I mean my G5 runs Commodore 64 software. OS9 will be a 7 year old OS by the time the PowerMac wintels ship.

What seven year old application do you have that you want Apple to support classic for??!! I don't think a 7 year old OS should be emulated...go buy a G4 Powermac (for cheap) and run it natively!

 

[RIP]

Do you really believe that? Just because a handful of enthusiastic developers say "oh this Intel box is sooo fast.." without giving any kind of hard numbers. While at the same time there are plenty of benchmarks showing that it takes the fastest available Xeon workstation (3.6 GHz) to be able to be a little bit faster and in some cases slower than a 2.7 GHz G5, testing the same real native apps on both systems.

What makes more sense? A bunch of unspecified claims, or real available numbers.

I wouldn't entirely discredit the developers. What current real world advantage or agenda would lying about the developer box performance benefit?

As it stands now, I tend to believe them however if I find out that any of these developers are on Apple's payroll, then I would rethink my position.

 

[rendezvouscp]

Why is it obvious that OSX wouldn't run windows apps? To me it makes sense from a practical standpoint, but is it feasible from a technical standpoint? Right, obviously OSX as it exists now does not run windows apps...but if the machine can run windows natively, why can't it run windows apps natively? I.E....what's to stop Apple from developing functionality into OSX 10.5 that will run .exe files? I suppose virii could become a problem were that the case, but you are saying that OSX could not run PC apps...but if OSX is built to run on PC hardware....and Windows is built to run on PC hardware...is it really so hard to he PC apps to run under OSX?

Technically? Yes. Actually possible? No.

Apple would need to figure out how to get .exe files to run. Then they'd need to figure out how to get .exe files to run under Mac OS X. Then they'd need to sort out the use of .exe files with the system. Then they'd need to figure out why the hell the went to all that trouble because it doesn't do anything to help them, only hurt them. Or maybe they should do that first.
-Chase

 

[aegisdesign]

I have not *seen* a developer box first hand, but this was clearly what was displayed on the one onstage at the keynote....

You don't need to see a developer box to see this. Download XCode 2.1 and the options are available there and PPC Mac owners can output universal binaries. Without an Intel Mac to test your code on, it's somewhat risky, but if you don't mind letting your users, which just now that means other developers only, test Intel compatibility then that's ok.

 

[ewinemiller]

Think about it. If Mac users can dual-boot Mac OS X or WinXP (or eventually Longhorn) Then what would be the point in software developers like Adobe, Microsoft, or (and especially) smaller developers & shareware developers making Mac-based software? Why make software for the 2% market share if that 2% can just buy a box copy of Windows & run their apps thusly?

A dearth in OS X-native software would seriously harm Apple's leverage, even if OS X is a better, easier to use OS.

I honestly hope Apple makes it as difficult as possible to load Windows on a Macintel box. (even though current speculation indicates they won't)

It would be a pity to see Mac OS X go down the same road IBM's OS/2 went.

I actually think you would attract more developers, especially the small ones, by having it dual boot. In my small niche market, many of my competitors don't bother with the Mac because the additional costs aren't really worth it until you hit a certain sales volume. However if you could buy just one machine that let you develop for both then that entry cost is reduced dramatically.

Also I don't think Windows compatibility was OS/2s downfall, it's more like Windows compatiblity was what kept it on life support. Nobody I know bought OS/2 for OS/2 they bought it because it really was a better Windows than Windows at the time. They were using it to run Windows apps exclusively.

 

[Durendal]

Assuming that this "article" isn't a load of tripe (which Appleinsider often posts), then it still proves nothing. There are no solid numbers, just a few quotes. Oooh, a friggin' web browser runs real fast! Gimme a break. Once the CPU hits a certain speed, the limiting factor is your connection. Also, Rosetta does NOT run Altivec code. When running plain vanilla, G3 code, I wouldn't be too surprised if the Intel Mac had an edge, but the G5 will still stomp the crap out of an Intel box when running Altivec. This article is useless.

 

[GFLPraxis]

Apple will be able to use SIMD (MMX <> SIMD - are they the same ?) which is the 'Altivec' for intel processors, so the loss of Altivec is no big deal.

IIRC MMX is nowhere near as good as AltiVec.

 

[RIP]

Technically? Yes. Actually possible? No.

Apple would need to figure out how to get .exe files to run. Then they'd need to figure out how to get .exe files to run under Mac OS X. Then they'd need to sort out the use of .exe files with the system. Then they'd need to figure out why the hell the went to all that trouble because it doesn't do anything to help them, only hurt them. Or maybe they should do that first.
-Chase

I can say with a relatively high level of confidence that a virtualization technology will be sold or even open-sourced by someone I would guess sometime around the time the very first Intel based Macintoshes are released into the wild.

We will be running windows apps on Intel Macs very much like we are running classic apps on current PPC Macs.

Mark my words.

The Macintosh will be taking the world by storm soon. And if I were the heads of MS, I would be very worried about their comments during the DOJ lawsuit they said about competition coming from nowhere when defending their monopoly in court.

 

[aegisdesign]

I'm sorry, but you don't know wtf you're talking about. A P4 3.6 is blazing, however you slice it. I have a P4 3.4 ghz with 1GB DDR2-533 memory, and it's fast as hell. It blows away any PowerMac when it comes to framerate at Doom 3, along with a slew of other tasks. Probably the only thing the PowerMac would be faster at is video encoding/decoding using Altivec. Integer performance, which most apps use, rocks on this thing.

Yawn... yet another PowerPC zealot spouting FUD about the platform.

Most PowerMac users care absolutely nothing about Doom 3 benchmarks and a whole lot about the performance of their video and graphics editing speed. And largely, PowerMacs soundly beat Intel based machines at Photoshop and video editing, especially if you compare Final Cut to After Effects/Premiere. This is where the speed matters to PowerMac buyers, not games.

The problem with Doom 3 isn't the processor speed, it's the speed of the graphics drivers and the non-optimal intel endian code in Doom 3. It just doesn't run well on the G5.

And I think if anything can be made from these reports, it's that when OSX is running intel optimized graphics drivers, the visible impression of speed - "Teh Snappy" - is more evident. This is because the majority of the work that goes into producing graphics drivers is to get optimal speed out of the Intel architecture, not PowerPC.

Otherwise it would be extremely hard to believe that somehow code running on OSX, compiled in GCC 4.0 for Intel has suddenly opened up a lead where a single Pentium 4 is running quicker than a Dual G5. Every single benchmark in the last few years has indicated you need a Dual Opteron or Xeon to get close to the G5 PowerMacs. And that's usually in benchmark wars with Windows where the MS and Intel compiler is a lot more efficient than GCC.

The anecdotes given of boot times, web pages 'snapping' to the screen as fast as IE6 on windows are simply useless. How often do you boot your Mac? Once a month? Once an update? and Safari 2.0 beats IE6 most of the time in my tests so that's not really a metric I'm impressed with.

Sorry, but it sounds to me like this is pure spin.

 

[ewinemiller]

IIRC MMX is nowhere near as good as AltiVec.

I think what they meant was SSE1-3. MMX was only integer, didn't do floating point. I've done SSE and Altivec code, Altivec was nicer, but not radically so. SSE3 is the latest and greatest and I think the only significant difference these days is that Altivec has more registers available, but with the right prefetching that's not really a problem.

 

[aegisdesign]

it is only a performance crisis if you compare a 1 inch thick PB with 4 hrs battery life to a 3GHz Intel 2.5 inch brick with 90 min battery life. essentially you're comparing a desktop replacement to a highly mobile machine.

Whilst there are still laptops out there like that, the Pentium M comes in at 2.1Ghz now and runs faster than those 3Ghz+ behemoths, gives better battery life than a PowerBook G4 and is equally as slim. The cheaper Pentium Ms down at the 1.6Ghz end will easily beat the G4 most of the time except for AltiVec enabled apps and those that use core graphics to offload a lot of the CPU to GPU.

1.6Ghz Pentium M laptops come in at prices below iBook prices these days so Apple is getting trounced there on speed and price. And they have no answer to the 2.1Ghz Pentium M.

That's where the Intel switch makes sense. It doesn't make so much sense on the desktop though. Not yet anyway and not with Intel's current roadmap for the desktop which really isn't that much more exciting than IBM's.

The switch isn't about CPU speed, it's about power and software.

 

[savar]

I disagree. It strains credibility to think that Apple has timed the Intel switch to correspond exactly to the moment when Intel will surpass the PowerPC architecture. It seems more likely that Apple exaggerated the PowerPC's superiority until the day when they announced switch to Intel, at which point they insisted on the superiority of the new platform. BTW, the whole "Performance units/watt" stinks of more Apple spin. What the hell is a perfomance unit?

The MHz myth hasn't died out at all, as some are claiming. In fact, Intel is finally admitting to it. Their whole roadmap is to start making chips which are more efficient. The Pentium M is laid out for energy efficiency, at the cost of max clock speed. This design philosophy is going to make its way into Intel's desktop processors as well.

And no, Apple isn't pulling this out of their butt. Performance/watt may be vaguely defined on stage for the public's sake, but there is a precise way of measuring that ratio in the lab. And the only reason why its only recently become a big deal is that the entire CPU industry is hitting the wall in terms of heat. Any move they make, i.e. smaller process, more transistors, only exacerbates the problem. Its finally time to design CPUs around the idea of efficiency because we simply can't push forward blindly anymore.

 

[aegisdesign]

Why is it obvious that OSX wouldn't run windows apps? To me it makes sense from a practical standpoint, but is it feasible from a technical standpoint? Right, obviously OSX as it exists now does not run windows apps...but if the machine can run windows natively, why can't it run windows apps natively? I.E....what's to stop Apple from developing functionality into OSX 10.5 that will run .exe files? I suppose virii could become a problem were that the case, but you are saying that OSX could not run PC apps...but if OSX is built to run on PC hardware....and Windows is built to run on PC hardware...is it really so hard to he PC apps to run under OSX?

To have them run seamlessly alongside OSX applications on your Finder...

a) You'd either need a much better version of WINE than currently is available and for it to be ported to MacOSX on intel and the Quartz APIs instead of X.

b) Or, you'd need a full copy of Windows installed like you do with VirtualPC

c) Or, you'd need a licensed windows compatibility layer underneath from Microsoft (like that will happen)

With each, you'd need Apple to figure out a way to make it integrate with the OSX Finder so cut, copy paste worked, drag and drop worked, printing, colorsync.... And then you've got two different windowing systems with different methods of use sitting onscreen at once. It worked on OS/2 because the GUI between windows and OS/2 was almost identical and WINE works on Linux as the GUI isn't that different usually between Linux and Windows. But MacOSX is completely different.

To have windows run in a window or fullscreen like VirtualPC does now from a real copy of Windows installed on a (virtual) disk partition would be easiest and with the virtualization technology coming in the next Intel chips, that's a whole lot easier. And that's what I reckon we'll see at best. A version of VirtualPC that runs at near 100% native speed.

Otherwise you'll have to dual boot.

I think seamless x86 Windows compatibility should not be part of the OS though. It'll be a disaster for parts of the Mac software industry, particularly games. What would be the point of writing a MacOSX version of a game if users can just boot into Windows? Or porting DirectX code to OpenGL for the Mac? It's the same reason the Linux games industry doesn't exist - Windows compatibility has killed off all but the most dedicated developers.

And I can see it affecting other sectors too. Most people work in mixed environments and offices where there may be Windows and MacOSX. I can just see it being rationalised by some PHB that since the Macs can now run Windows, they'll only buy the Windows site licence. And then those windows versions get leaked off home by the employees and nobody bothers buying Mac software.

 

[Peyote]

Technically? Yes. Actually possible? No.

Apple would need to figure out how to get .exe files to run. Then they'd need to figure out how to get .exe files to run under Mac OS X. Then they'd need to sort out the use of .exe files with the system. Then they'd need to figure out why the hell the went to all that trouble because it doesn't do anything to help them, only hurt them. Or maybe they should do that first.
-Chase

My thinking is...what if this is the beginning of the end for two independent platforms? Imagine if you could run any PC or mac software on your machine...natively (somewhat at least). No more need for higher prices for mac software. No more need for porting anything to the mac. No more need for running Windows to use PC apps. No more need for Windows.


"Why buy a PC when a Macintel can do it all?"

Honestly, why would it be such a threat to Apple anyway? They're already going to have PC apps running on macs through Windows...why is it that THAT isn't a threat, but PC apps running through OSX is? And how will it hurt Apple anyhow?


I hear your arguement about why Apple would not be able in a practical sense get PC apps to run on OSX...but I don't hear any technical reasoning behind it, just that "they would need to figure out how to get .exe files to run under OSX"....well why is that hard? They figured out how to run OSX on a PC...why would it be so hard to get PC apps to run on OSX? I'm not trying to be arguementative...I'm really honestly curious...give me a technical explaination and I'll say, "ok I understand why that is not practical."

 

[Peyote]

To have them run seamlessly alongside OSX applications on your Finder...

a) You'd either need a much better version of WINE than currently is available and for it to be ported to MacOSX on intel and the Quartz APIs instead of X.

b) Or, you'd need a full copy of Windows installed like you do with VirtualPC

c) Or, you'd need a licensed windows compatibility layer underneath from Microsoft (like that will happen)

With each, you'd need Apple to figure out a way to make it integrate with the OSX Finder so cut, copy paste worked, drag and drop worked, printing, colorsync.... And then you've got two different windowing systems with different methods of use sitting onscreen at once. It worked on OS/2 because the GUI between windows and OS/2 was almost identical and WINE works on Linux as the GUI isn't that different usually between Linux and Windows. But MacOSX is completely different.

To have windows run in a window or fullscreen like VirtualPC does now from a real copy of Windows installed on a (virtual) disk partition would be easiest and with the virtualization technology coming in the next Intel chips, that's a whole lot easier. And that's what I reckon we'll see at best. A version of VirtualPC that runs at near 100% native speed.

Otherwise you'll have to dual boot.

I think seamless x86 Windows compatibility should not be part of the OS though. It'll be a disaster for parts of the Mac software industry, particularly games. What would be the point of writing a MacOSX version of a game if users can just boot into Windows? Or porting DirectX code to OpenGL for the Mac? It's the same reason the Linux games industry doesn't exist - Windows compatibility has killed off all but the most dedicated developers.

And I can see it affecting other sectors too. Most people work in mixed environments and offices where there may be Windows and MacOSX. I can just see it being rationalised by some PHB that since the Macs can now run Windows, they'll only buy the Windows site licence. And then those windows versions get leaked off home by the employees and nobody bothers buying Mac software.

That's a valid arguement

 

[aegisdesign]

And no, Apple isn't pulling this out of their butt. Performance/watt may be vaguely defined on stage for the public's sake, but there is a precise way of measuring that ratio in the lab.

And I'm sure the ratio isn't 15 units for PowerPC and 70 units per watt for Intel as Steve quoted. Which means he's either spinning like mad or Intel have something up their sleeves that isn't in their roadmap.

Or perhaps he was meaning as a complete platform and he was comparing a Pentium M with Intel Xtreme graphics to a PowerPC G5 with an nVidia 6800.

 

[iMeowbot]

I think what they meant was SSE1-3. MMX was only integer, didn't do floating point. I've done SSE and Altivec code, Altivec was nicer, but not radically so. SSE3 is the latest and greatest and I think the only significant difference these days is that Altivec has more registers available, but with the right prefetching that's not really a problem.

Yeah, that's about the size of it. Word is that the first production Macintels will only have SSE2 available, but there are platform-independent vector functions (and nice clean ways to tell what is available for those who insist on writing the stuff by hand) so that it should scale up to the zoomier stuff cleanly.

 

[kzg]

Wow, there's so much to this transition that my mind is almost boggled. There's a few things that I think are interesting though.

Right now, the PPC kicks ass at math-related computational problems. I believe that's the main reason why there are so many fantastic clusters using the G5. However, we're losing that with the Intel switch, which makes me wonder why Apple would go ahead and lose something like that for their customers. The transition is supposed to make things faster, not same or worse.

-Chase

Unfortunately the PPC isn't keeping up, of the top 500 supercomputers in the world only one is apple based.

 

[Stella]

Windows and OSX are *completely* different beasties. Just because they run under the same processor doesn't make software compatible.

The closest you'd come to it would be running WINE under OSX.

IF you could run windows apps under OSX software companies would be less inclined to port their apps to OSX.. a lot less market share so why invest the time when Mac users could run Windows apps anyway?

Running Windows natively under OSX is a very real threat for Apple.

History has shown this: Take IBM OS2:
IBM's OS2 ( Pre Warp )could run Windows apps ( better than Windows ;-) ) and of course, native Apps. A lot of companies didn't bother writing native apps for OS2 because users could run windows anyway.
( No difference between this example and OSX v Windows )

Apple are taking the correct approach - no native Windows apps under OSX. Dual boot or emulation only.

I hear your arguement about why Apple would not be able in a practical sense get PC apps to run on OSX...but I don't hear any technical reasoning behind it, just that "they would need to figure out how to get .exe files to run under OSX"....well why is that hard? They figured out how to run OSX on a PC...why would it be so hard to get PC apps to run on OSX? I'm not trying to be arguementative...I'm really honestly curious...give me a technical explaination and I'll say, "ok I understand why that is not practical."

 

[dejo]

Unfortunately the PPC isn't keeping up, of the top 500 supercomputers in the world only one is apple based.

Just because it's not Apple-based, doesn't mean it's not PowerPC-based. For example, the world's fastest supercomputer in a single installation is IBM's Blue Gene/L prototype and uses PowerPC 440 700 MHz processors.

Top 500 Supercomputer Sites

 

[Krevnik]

What's the difference now? They switched from two baskets to a different two baskets.

Yes, but here is the thing, the design of the OS is such that they could switch back to the first two baskets in the future if they wanted to, and to an entirely different two baskets that we don't know about. (Although there is really only 3 viable consumer Archs right now: x86(-64), PPC, and MIPS)

 

[Krevnik]

I think though, that due to the lack of AltiVec, the Intel Macs won't have QUITE the same Photoshop performance. PC's struggle a bit more on that generally. But the PC's usually do simple stuff like web browsing a lot faster.

If Adobe optimized for SSE2/3, then we wouldn't see the hit. Apple might be able to convince them to do that (and Adobe gets a boost on Windows from it too).

 

[pubwvj]

Yeah, I'm particularly excited how much faster the browsers will be under OS X-Intel.

Interesting statement. I'm curious. Why? Browsing on my old 500MHz G3 PowerBook Pismo is fast under Safari and OSX 10.3.9. The limiting effect on browsing speed is 1) connection speed (I have DSL) and 2) server speed on the other end (OSX-Intel will have no effect on that).

 

[Multimedia]

January 10 '06 SteveNote May Introduce Dual Core Yonah PowerBooks

Just read over at AppleInsider that the speed of Apple's Intel Developer systems are impressing the developers, even when PowerPC programs are running under Rosetta.

With this latest bit of news, I think I just might wait until they release a Intel-based PowerBook before buying one. Who else here is planning to do the same? Anyone think this would be a bad idea? And, when do you guys/gals think we'll see the first Intel-based Macs, specifically the PB's.

I am still using first gen October '03 1.25 GHz 15" G4 AlPB. If they ship a dual core PPC G5 PowerBook, I will buy that. But not if it's only 1.6 GHz single core. In that case, like you, I'll wait for a dual core Yonah PowerBook which I expect will happen early next year IF Intel can overcome it's Yonah production problems by Fall '05 and crank it up. I'm hoping for a dual core Yonah PowerBook announcement during the SteveNote on Tuesday January 10 '06 at MacWorld San Francisco.

 

[illumin8]

Every single benchmark in the last few years has indicated you need a Dual Opteron or Xeon to get close to the G5 PowerMacs. And that's usually in benchmark wars with Windows where the MS and Intel compiler is a lot more efficient than GCC.

This is the whole point I'm trying to get to. Apple distorts benchmarks. They have done this for years. Sure, I can show you a benchmark of Altivec-optimized photoshop plugins where a G5 absolutely spanks any P4. But the truth of the matter is that for almost all purposes, whether it's web-browsing, document creation, listening to iTunes, pretty much anything but Video/Audio encoding, the P4 is faster.

I should point out that you never qualified your original argument by saying "after 6 months to a year when the registry gets bloated". Everyone that uses Windows knows this; that's just a limitation of a poorly designed OS.

But it will be fun to see all of the PowerPC zealots eat crow when the first dual-core Yonahs hit the market and are benchmarked against PowerPC products running the same apps.

I am reminded of the following great quote:

"I have never understood why it should be necessary to become irrational in order to prove that you care, or indeed, why it
should be necessary to prove it at all." -Kerr Avon, "Duel"