Macintel

[amichel]

Well, if no one else is gonna post it, I'll ruin everyone's day. Steve officially announced Apple's switch to Intel today at WWDC.

Brilliant or buggered?

Discuss.

[lnp]

Brilliant. Why? Because I think its 90% done in good faith and 10% bluff. It might light the fire in IBM to develop a cooler version of the G5 (small probability) or it'll mean that Intel does it and the powerbook line will become a real complement to the new desktops. The only risk is the necessary recompilation of the software (I only know vaguely what this means, could someone explain it better?) and the lag in sales that will result as people wait for the new jump to be made.

But yeah. I think a good deal all around.

[amichel]

Compilation. Well, lets start with what a compiler does, just to be sure we're on the same page. Compilers take a high level language, like Objective-C in Apple's case, and translates/optimizes it into binary instructions specifically for your computer's central processor. Different processor architectures (x86, PowerPC, Sparc, Alpha, etc) have different instruction sets, different registers available for different things, and generally different aptitudes, which means that code compiled for one simply won't run on the others. Recompiling is just taking your source code and compiling it for a different CPU. The trick is, especially with something like PPC to x86, if you've been a REALLY industrious programmer and gotten in there with some low-level optimizations of your own in your code, or even just written high-level code that takes advantage of specific hardware features of your current architecture (read Altivec on PowerPC), you now have to hack a little to get that code to be executable on the new architecture. The Steveinator is saying that if you wrote Cocoa (native OS X) apps the right way from day one, you'll have only a few minor tweaks to make in your source code and it'll be Macintel compatible, which is good news for developers who write architecture-independent code, a growing trend. Carbonized apps (OS9 apps that were only slightly modified to make them run natively in OSX) take a bit more work to get away from OS9-isms, but according to Steve are still easily portable with a small time commitment. Overall, I think Apple has considered the developer transition pretty thoroughly, and as such has already addressed many issues that would have given developers the willys.

One thing of note is that Apple has struck a deal with Intel to include the Intel C++ Compiler with Xcode, as well as the optimized Intel libraries. For the uninitiated, the Intel C++ Compiler is an alternative to GCC (the GNU C Complier, an open-source project) when compiling code for Intel chips, and it tends to produce noticably faster code since it contains all sorts of scary Pentium-only optimization routines. This is good for Apple developers because their code will compile into beautifully blazing bits of binary badassedness.

Now that the lesson is over, as for the 10% bluff, I think no. IBM and Moto have consistently shown that they have no interest in focusing their research and development efforts on keeping PowerPC in the desktop market. The PPC970 (G5) was a stopgap measure to keep Apple on board as a PowerPC PR darling, while IBM fired up the "Power Everywhere" campaign. Now that Power chips are running EVERY next-gen game console, they've got all the media exposure they need for PPC, and Apple is simply vestigial, and would probably have been consistently ignored until it went away anyway. The 970 can wither and die, much to IBM's general delight, and they can focus on what they consider the future of Power, the Cell, which is a dubious proposition, but one they seem committed to. By taking on Apple as a high-profile (low-volume, but still) client, Intel gets to send a quiet-riot-style flip-the-bird over at Microsoft, who decided that IBM was the place to be for the Xbox360, despite the years of mutual support given and received by Intel. Once MacOS runs on Intel hardware, Intel can finally get behind an OS partner that isn't being crucified once a week for yet another catastrophic security flaw, and will finally show what x86 can do with a crack team of quality-conscious software engineers who are used to and good at squeezing every benefit they can out of an architecture. The double-bonus for Intel is that the software comes hand-in-hand with the best industrial design team on earth, who will finally make the Pentium sexy, unlike Dell and its ilk who have done little except replace beige with greyish-black. Mac users get to keep their spiffy trademark looks, and since most don't know a G5 from a redwood tree, all they'll notice is that Macs are way-the-heck faster than they used to be. Developers get to mull the mouth-watering possibilities of HD in the living room, as well as easy porting between OSX and Windows, as I'd imagine Apple has some tricks up its Developer Connection sleeve that will make porting Windows apps much easier. Intel gets good press, Apple gets a price/performance ratio that makes the Mac a much more viable PC-replacement, and they get the added benefit of being on the same architecture as the PC. This allows big-ticket development houses that have traditionally shunned the Mac entirely (read Autodesk, the CAD kings who also own Discreet of 3DSMax fame) to reconsider the possibility of putting their products on the Mac, now that a port is little more than a quick re-compile using the OSX shared libs, with free access to the Intel C++ compiler and Xcode. This gives them access to a whole new market of content creation and design professionals who have been alienated by their platform choice. Mac design pros account for big money as far as workstation dollars go, and I doubt these companies will pass up the opportunity to tap into that cash-flow.

My personal predictions for the whole product line go something like this:





PowerMacs

Dual Xeon processors
Probably dual-core if not quad-core by the time Apple is using them
Implication: 8 logical processors for super parallelism, and big wins in multimedia apps
Prices remain almost the same, with the possible introduction of an even higher-end model since chip supplies will be virtually unlimited. Quad-processor quad-core boxes are not infeasible. 16 logical processors means real-time real-fast rendering. Scary.




Powerbooks

Single Intel Pentium-M (Dothan-derivative core) at 2-3Ghz
Again, Probably dual-core by the time Apple gets to them
Implication: dual-cpu top-tier performance from your portable
Powerbooks regain their place as viable content creation/delivery platforms
Pricing remains consistent with drops on the low end, but performance boosts across the board




iMacs

Single, possibly dual Pentium-M (Dothan again)
Probably dual-core
Better cooling performance on iMac allows Dothan processors to really show their muscle
iMacs become true next-generation multimedia viewing consoles, allowing multimedia pros delivering in Quicktime to deliver H.264 encoded 1080 HD content to the consumer user-base
Enclosures that are even thinner and more awe-inspiring without the G5 creating a volcanic heat-wave




Mac Mini

Single Dothan (probably around 2GHz)
Probably dual-core
Add a Blu-Ray drive
Mac mini becomes a perfect media center with a chip that can turn out H.264 at HD resolutions without a hitch
Steve finally gets to infiltrate your living room, like he always wanted




iBooks

Single Celeron-M
Not as bad as you'd think, Celeron does not indicate the sort of performance black hole traditionally associated with the branding. Celeron-Ms are kid brothers to the Centrino chips (Pentium-M Banias and Dothan) and are basically cache-crippled and will probably be saddled with slower system busses and chincy graphics cards, much like the iBooks we have now.
The difference is, the low-end model will be more like $500 dollars instead of $1000.

With difficulty, I second the vote of brilliant. Don't get me wrong, I'm still slightly queasy at the thought of Intel-inside my Mac, but I can see the benefits for everyone involved, including us as consumers and developers. I think once the shock wears off, I'll allow all the good news I've just evangelized to truly sink in. Then I can stop trying to rebuff The Steve's RDF and get back to the business of being a Mac Zealot.

[abudhu]

Brilliant. I have been discussing this on other forums, though our conversations have derailed into enthusiast AMD fans (such as myself) feeling sad that AMD got the shaft from Apple. Though the arguement put forth about AMD not being able to meet production quanties is valid it still sucks. But hey, as Apple is going x86, nothing is stopping us from sticking an AMD chip in there instead. Well, actually there is such as Apple's different BIOS but we won't get into that.

Vote of Brilliance.

Can you imagine Tri Booting: Windows/Linux/Mac
That is so what I would do. Or just install Mac OS onto VMWare.

*Wink* Everything in Adams first paragraphs reminds me of PearPC no?

[jwfische]

It'll just be weird seeing that "Intel Inside" sticker slapped on my future Mac.

[fischerm]

Brilliant.

I think the whole iPod phenomenon has given Apple a taste of what it's like to be the dominant player in a market, and they liked it. Apple has had all sorts of supply issues with IBM and Moto over the years, and Intel has capacity to spare.

Consider a few things:
Apple has said that they will do nothing to prevent Windows from running on mac hardware, so Amit's post about Mac/Linux/Windows all on one box certainly seems likely to happen. (they will of course take great pains to prevent you from running MacOS X on non-apple hardware)

Things like Virtual PC will be much much faster.

Wine (http://www.winehq.com/) should run on the new macs, meaning you may not even need Windows.

All the graphics card/sound card/blah card people will have less R&D to do if Apple adopts Intel's reference designs for motherboards, so Mac users are more likely to be treated as first class citizens.

There will be tons of work ahead, but we live in interesting times.

[abudhu]

Wine (http://www.winehq.com/) should run on the new macs, meaning you may not even need Windows.


WINE (under linux) uses Windows files and keys and reattaches them to allow programs to work. So, you still need the needed files, which are sometimes found in the WINDOWS directory. This is why when trying to run, lets say Photoshop 7 under Linux, WINE creates a FAKE Windows System DIR with DLLS and such. It needs to find the connections and remake them in order to work.

On a side note, getting Photoshop CS to work under WINE is a Pain in the ARSE!

[emurphy1]

Well, being here at WWDC I can tell you it has been interesting. Suddenly, late yesterday afternoon there were all sorts of changes to the sessions schedule as the "big secret" was finally out. Today developers were rushing to sessions on XCode 2.1 (just released here and needed for creation of "Universal Binary"). There is a session later today on porting your applications to Intel based OS X, and there is even a Universal Binary lab full of the same P4 3.4 Ghz machines (with sweet 30 inch flat panels) that Jobs used in yesterdays keynote. The idea in the Universal Binary lab is you bring in your code, used XCode 2.1 to compile for PPC or i686. Pretty, cool, easy to do...but as Adam has stated the code needs to be pretty netural to start with. The example, I worked with Apple Engineers on involved downloading the source for a solitare game from versiontracker.com and then compiling it. Note: the solitare game is a Cocoa app.

I'm not sure about including Intel's C++ compiler though. All I've seen here is talk about GCC 4.0 and how to use GCC 4 as the OS X compiler of choice.

The talks I've been in with other developers in the hall, or at lunch seems to point to two big factors for the Apple-Intel marriage.
1) IBM is swamped (and happy) with making chips for XBox consoles. Plus IBM promised a 3+ Ghz chip a year ago to Apple and has never delivered.
2) The power usage and heat factors of IBMs PPC chip are big problems/worries for Apple. Intel has solved both those problems with its P mobile line.

Mark and Amit are right...you may, just maybe, be able to turn your new Mac into a dual boot Linux/Windows/OS X machine...but you will never (or at least if Apple has its way) run OS X on a Dell/Gateway box. As for Wine...yuck, yuck, yuck...my experience with trying to run Windows apps via Wine on Linux is bad. Total pain, doesn't work right, features are missing...horrible mess, not worth the hassle...just use Open Office or Gimp instead.

[abudhu]

Yeah, WINE isn't too fun to use. Though its quite amazing when you *do* get something to work in it :)
Not too pleased with CUPS as well. Stupid thing won't detect my Printer, even with the right "drivers" installed.

[amichel]

Just so you guys know I'm not high on crack, Apple announced the ICC compiler in their own press release here (http://www.apple.com/pr/library/2005/jun/06intel.html).

A quote:
"Intel plans to provide industry leading development tools support for Apple later this year, including the Intel C/C++ Compiler for Apple, Intel Fortran Compiler for Apple, Intel Math Kernel Libraries for Apple and Intel Integrated Performance Primitives for Apple."

GCC is good, but for x86, ICC is f'n metal.

You can already dual boot any mac you like with Linux/BSD and OSX, so dual booting a mac isn't really a new thing. Booting windows, well, that's pretty huge. But I think the more important implication of an x86 migration is that you may never NEED to boot windows again. All the programs that are windows specific, possibly even a good number of games, just became much easier to port. I'm not hoping that Apple lets me boot windows, I'm hoping Apple just negated my NEED for windows altogether.

[abudhu]

However, there are currently...3 Distros that are Mac compatible? Ubuntu, Yellow Dog, and I believe Knoppix has one. None of these distros are to "hot" as well. In anycase, as for PC games being ported to Mac when they are x86, I am willing to bet Big Bad Bill Gates, will introduce some proprietary code or what not then use his power to influence the little/medium game companies to add it to their code that will only allow it to work on XP+.

You wait and see, I am sure Bill Gates is 1) not to happy about this 2) Will do whatever he can to screw All Mac lovers over.

Linux almost pulled me all the way over from Windows...except when my Printer started printing 1 page per every 5 minutes. :-\

[abudhu]

Thought you all would like this:

http://www.apple.com/quicktime/qtv/wwdc05/

Its pretty darn long but well worth it. They have been recompiling OS X since 10.0, for INTEL. Thats insane! The whole Xcode part was pretty funny as well.

Why can't more people just love Both PC's and MAC's like I have come to do? :-D

[amichel]

However, there are currently...3 Distros that are Mac compatible? Ubuntu, Yellow Dog, and I believe Knoppix has one. None of these distros are to "hot" as well. In anycase, as for PC games being ported to Mac when they are x86, I am willing to bet Big Bad Bill Gates, will introduce some proprietary code or what not then use his power to influence the little/medium game companies to add it to their code that will only allow it to work on XP+.

You wait and see, I am sure Bill Gates is 1) not to happy about this 2) Will do whatever he can to screw All Mac lovers over.

Linux almost pulled me all the way over from Windows...except when my Printer started printing 1 page per every 5 minutes. :-\
3 distros? Jeeze, lets straighten that out right now. Ubuntu, Yellow Dog, Debian, SuSE, Gentoo, Slackware (via the Slackintosh port), Fedora... And you could always roll your own LFS if you felt the urge. I'm sure there are others I've left out. I swear sometimes you just say things to make me angry. That was just silly. Did you even Google that insanity? I'm not even mentioning the BSDs.

As for your "hot" moniker, I don't know how you go about determining if distros are "hot", but YellowDog is quite capable and works about like RedHat in the general sense, not to mention their totally sweet BlackLab distro that lets you cluster Macs on the easy. As for Ubuntu, I've only heard good things. I guess I don't know what you consider "hot", but whatever it is, it's apparently not popular. I'd wager that you've probably not even tried a single distro on a Mac, though it's possible you just gave up too quickly. Now that your ignorance has been alleviated as to the number of popular distros available to try, perhaps you can give it another shot.

On to the games. I guess maybe if you don't think about it, what you said makes sense. Fact is, Bill needs game developers. Not the other way around. Most game development houses don't lean on PC gaming as their sugar-platform, they lean on consoles. Much like a highest-end GPU is for show, and a mid range is for dough, consles are the bread and butter of the gaming industry. If Bill tries to screw them, they'll walk. People will buy consoles. They'll still be in business. But without games, Windows loses a serious edge in the marketplace. MS won't chance that. Bill can't lock anyone to anything. All he can do is try to make sure Windows stays the most viable development platform by keeping it technologically choice for gaming, which, at the moment it is, thanks to DirectX. However, as we've seen with id software, if you write clever OpenGL code, the Mac is quite viable, Doom 3 as evidence. Once ATi and Nvidia are able to port some of their x86 platform driver optimizations over to the Mac, I'd say we stand to see favorable improvements to GPU performance. My point was that there are serious hurdles to Mac porting that are removed by platform consistency. It's now a matter of writing an agnostic OpenGL engine and compiling against two sets of libraries rather than reoptimizing your entire source tree to take advantage of a different architecture. OpenGL is OpenGL everywhere. So, time will tell on Mac games, but Bill has nothing to do with it.

And frankly, Bill doesn't have to be happy about the switch, because the only thing he has to dangle over Mac users is Office. Period. And judging from things like Keynote and Pages, I'd say Apple is probably on the verge of an Apple-branded office productivity suite, and even if they aren't, OpenOffice.org 2.0 is a quantum leap forward for open source office, and with projects like NeoOffice/J porting to the Mac, it's only a matter of time before alternatives like OpenOffice, AbiWord and others are recognized as viable replacements. I wouldn't put it past Apple to hop behind yet another open source project and develop AppleOffice on the OOo tree, just like webkit on the konq tree. They've seen the benefits of working with open source, again and again. They know it's there if they want it. And Microsoft knows it too.

CUPS is great, but it certainly takes some effort. If I were you, I'd start here (http://www.linuxprinting.org/printer_list.cgi) and make sure your printer doesn't have a better driver method. You might be doing it the hard way.

At times, I am truly baffled by your posts, Amit. This being one of them.

[abudhu]

That website: Been there done that.

Your Linux Distros: Get YOUR Facts straight. Find me the PowerPC SuSE and I'll shutup.

http://www.frozentech.com/content/livecd.php?pick=Linux_PPC&showonly=All&sort=&sm=1

Sorted for PowerPC. Please note that though the name may be LiveCD if you take a look at the PRIMARY function you will see what its used for. Desktop replacement being a fully functional installable OS.

---
Edit: Conceded point to SuSE. Though the article I read had it for recompiling SuSE 7.1 to PPC. But that is a lot of effort.

[amichel]

Fair enough:
http://www.penguinppc.org/about/distributions.php

Do I need to click them all for you and post links, or can you take care of that bit yourself? Certainly you can read the "SUSE Linux Enterprise Server" bit at the bottom, no recompile necessary.

[abudhu]

Hm. Neat. But do you really have the money to buy the Enterprise Server, I know I don't. :( Nor would anyone outside a business need a Server edition of Linux. So, I partly still stand it that niether, in this case, SuSE, or Mandriva Standard/Professional Editions will run on Mac.

As for your attacks on me saying "hot", they are misplaced. I believe by "hot" I meant, a major distribution. Distro's such as Knoppix are not considered as major distributions for their sometimes lack of features. As for trying Linux on Mac I have used LiveCDs on Mac's before...and wow...they operate as a Linux machine should. Tried Both Ubuntu and Gnoppix.

I assume you have Benchmarked the OpenGL DOOM 3 mac? Head over to HardOCP, and try to find the article if you can. The Mac port of it using OpenGL falls heavily to Direct X.

As for my Printer, well its a HP Laser Jet Series II...made in 1990? No Printer page has ever listed the drivers as stable working under it. As I had said...it prints...just not very fast.

Btw: What in the world is a "New World Mac" What series does this refer to?
--
Edit:
Did a bit of snooping
ftp://ftp.net.usf.edu/pub/fedora/linux/core/development/ppc/
I can't seem to find the .iso for the install. Help.
--
Edit 2:
In anycase I apologize. This thread was about MACINTEL not MACLINUX.

Xcode 2.1 seems darn cool. But they priced a .X revision at 1000 dollars! Wow! Seems a bit pricy. Thus this simple point brings me to why I have been put off by Apple, Prices. Could their systems be any more expensive? Insane. The CEO Of WolfRam Research was crazy.

I don't really like how they introduced the widget system to Tiger. Do you really have to keep calling up that deskport background app to see all your active widgest? Is there a way, like Konfabulator, where you can actually have your widgets on the screen while you surf the web/write papers? Every time in that Movie, he would bring up all the widgets, the background would dim, and the widgets appeared. That seems like a waste. Who would want to use a widget for cnn news by having to bring up another app (albeit a very very quickly loading background app) when they could just launch safari?

[amichel]

Xcode 2.1 is freely downloadable from the ADC. The $999 is for a lease on a pre-production Intel Mac so you can start porting your software now and test your apps on Rosetta. Xcode is and will always be free, according to Apple.

The widget system is fantastic in my opinion. I use dashboard all the time. I used konfab before, but there were only a few konfab gadgets that I found useful to have open all the time. Konfab also introduced a dashboard-style interface a little while ago that does exactly what Dashboard does. Widgets aren't meant to be used as full-on apps, though some people may write them that way. They were supposed to be small, lightweight mini-programs that were easy to code and provided quick access to often needed information like phone numbers, current weather, stickies, etc. It's just to get those little mini-app windows out of your way so you can use your screen realestate for real apps like Word or Photoshop. Think of it more as a secondary virtual desktop. I use my dashboard all the time. I think once you have it and get a few widgets in hand, you get an idea of what it's good for.

In response to your question though, there IS a way to get them off the dashboard onto the desktop, though it's not an officially supported thing for end users and the widgets cannot be hidden or level-switched, meaning that they will always be on top of everything in their area. It's a hack using developer mode, and the process is documented here (http://www.macosxhints.com/article.php?story=20050422172929402) at macosxhints.com. They have loads of hints dealing just with dashboard if you're looking for other clever ways to use the dash. Some of my favorites are:


At a glance Dashboard (http://www.macosxhints.com/article.php?story=20050524154252310)
Google Maps in the Phonebook search (http://www.macosxhints.com/article.php?story=20050424065456133)
Slow Motion Dashboard (and other things) (http://www.macosxhints.com/article.php?story=20050503101357824)
Personalizing the World Clock (http://www.macosxhints.com/article.php?story=20050430193250534)

[tsubidio]

Can you imagine Tri Booting: Windows/Linux/Mac
That is so what I would do. Or just install Mac OS onto VMWare.

Sorry for going OT, I just want to say that this is already possible, as the pictures below (taken a few minutes ago by myself with my cell) are witnessing.

http://img499.imageshack.us/img499/1483/dsc003505bf.th.jpg (http://img499.imageshack.us/img499/1483/dsc003505bf.jpg) http://img498.imageshack.us/img498/7558/dsc003520gn.th.jpg (http://img498.imageshack.us/img498/7558/dsc003520gn.jpg) http://img492.imageshack.us/img492/8173/dsc003555ch.th.jpg (http://img492.imageshack.us/img492/8173/dsc003555ch.jpg)


Of course it is not 100% working, mostly just because my Pentium M does not support SSE3 but only SSE2, and therefore I had to use a patch to basically "emulate" SSE3, so that PPC applications would work. Moreover my ATI Radeon graphic card is not supported and I can only use one resolution (1024x768). The whole thing would be solved if you have a more modern CPU that supports SSE3 and an NVidia graphic card (I'm pretty sure it's supported, but I would have to look it up to be absolutely positive).
If you wonder how I got it to work, I made an image of the HD of my PowerBook (running OS X Tiger), and followed tips found on this website: http://wiki.osx86project.org/wiki/index.php/Main_Page

PS: click on the image for a bigger view. Sorry about quality, but if there is no light my cell's camera doesn't work too well.
Edit: I know, the time on OS X is wrong, I just never really cared, I mainly installed it to see if it actually was possible to do it, and what were the restrictions compared to a regular mac. I still only use Linux for most of the things I do, except for when I'm playing videogames :P
Oh, btw, speed on "OSx86" is great.
Edit2: I really didn't notice how old this thread was, sorry about this, I'm sure that by now all of you know this already :(

[abudhu]

Matt:
I did it already as well, I was just too lazy to post any Screen Shots. It was fun.