Slashdot Mirror
Apple Switching to Intel
Steve Jobs announced at the WWDC keynote today that Apple is switching to Intel processors. MacNN has live coverage. The bottom line is that Mac OS X for the last five years has been running on Intel, the switch is expected to be complete in two years, and Rosetta will allow PPC apps to run on Intel-based Macs, transparently. If you're using Xcode, it is small changes and a recompile; otherwise, you might be seeing a lot of work ahead of you. You will be able to order the 10.4.1 preview for Intel today.
The rumors are true: Intel will be inside
Jobs talked about the major transitions in the Mac's life -- starting from the Mac's Motorola 68000-series processor to PowerPC. "The PowerPC set Apple up fro the next decade. It was a good move," he said.
"The second transition was even better -- the transition from Mac OS 9 to Mac OS X that we just did," he continued. "This was a brain transplant. And even though these operating systems (9 and x) vary only by one in name, they are very different, and this has set Apple up for the next 20 years."
As the Intel logo lowered on the stage screen, Jobs said, "We are going to make the transition from PowerPC to Intel processors, and we are going to do it for you now, and for our customers next year. Why? Because we want to be making the best computer for our customers looking forward."
"I stood up here two years ago and promised you 3.0 GHz. I think a lot of you would like a G5 in your PowerBook, and we haven't been able to deliver that to you," said Jobs. "But as we look ahead, and though we've got great products now, and great PowerPC products still to come, we can envision great products we want to build, and we can't envision how to build them with the current PowerPC roadmap," said Jobs.
Intel processors provide more performance per watt than PowerPC processors do, said Jobs. "When we look at future roadmaps, mid-2006 and beyond, we see PoweRPC gives us 15 units of perfomance per watt, but Intel's roadmap gives us 70. And so this tells us what we have to do," he explained.
Transition to Intel by 2007, and yes, Marklar exists
"Starting next year, we will introduce Macs with Intel processors," said Jobs. "This time next year, we plan to ship Macs with Intel processors. In two years, our plan is that the transition will be mostly complete, and will be complete by end of 2007."
Jobs then confirmed a long-held belief that Apple was working on an Intel-compatible version of Mac OS X that some have termed "Marklar."
Mac OS X has been "leading a secret double life" for the past five years, said Jobs. "So today for the first time, I can confirm the rumors that every release of Mac OS X has been compiled for PowerPC and Intel. This has been going on for the last five years."
Jobs demonstrated a version of Mac OS X running on a 3.6GHz Pentium 4-processor equipped system, running a build of Mac OS X v10.4.1. He showed Dashboard widgets, Spotlight, iCal, Apple's Mail, Safari and iPhoto all working on the Intel-based system.
Apple needs developers' help to complete the transition
"We are very far along on this, but we're not done," said Jobs. "Which is why we're going to put it in your hands very soon, so you can help us finish it."
Widget, scripts and Java applications should work in the new environment without any conversion, said Jobs. Cocoa-based applications will require "a few minor tweaks and a recompile." Carbon-based applications require "a few more tweaks," recompiling, and "they'll work," said Jobs. And projects built using Metrowerks' CodeWarrior need to be moved to Xcode.
The future of Mac OS X development is moving to Xcode, said Jobs. Of Apple's top 100 developers, more than half -- 56 percent -- are already using Xcode, and 25 percent are in the process of switching to Xcode. "Less than 20 percent are not on board yet. Now is a good time to get on board," said Jobs.
A new build of Xcode, version 2.1, is being released today. This new release enables developers to specify PowerPC or Intel architectures. "... and you're going to build what's called a universal binary. It contains all the bits for both architectures," said Jobs. "One binary, works on both PowerPC and Intel architecture. So you can ship one CD that supports both processors."
"This is nothing like Carbonizing"
Many developers reading this news may be thinking that they'll have to go through the same woes they had to in order to get their Mac OS 9 applications "Carbonized" to run on
Here's Apple's press release.
Dispel any remaining doubts; we are now living in the evil mirror universe.
Possibly not - the new version of XCode builds universal binaries for both Intel and PPC. So, what's the problem again?
You have plenty of time. The rumors were only half-true.
Apple is adopting Intel, but is not "ditching" IBM.
New G5 towers will still be around for at least another year, and probably at least two. Intel is probably going to start by replacing the G4 CPUs in Powerbooks and minis.
At the Stevenote, he informed devs that they would be supporting both platforms for a long time to come.
Information wants to be anthropomorphized.
1. If Rosetta works as well as demonstrated (Jobs showed unmodified PPC versions of Photoshop+filters and MS Office running happily and fast on the Intel Mac box) then this will be less painful than you think.
2. The way the Intel and PowerPC raodmaps are going I think in three uears time there will be a HUGE difference in capability. Jobs was demoing a Pentium 3.6GHz quad for God's sake!
Do you have any evidence to back this assertion? Generally speaking, Altivec in the G5 has the same function and performance as SSE2 in the Pentium 4. I use floating point functions that I have developed and coded in assembly language myself, and I don't see any difference between Altivec and SSE2 at the fundamental level.
Most of the derogatory comments by Apple users about the supposed shortcomings of SSE2 are ill informed, they seem to confuse SSE2 with MMX. Optimization for either the Altivec or SSE2 is a complex subject. First, one has to find an algorithm that works well for vector operations, which means making sure that add and multiply operations will overlap correctly. Then one has to adapt that algorithm for the cache size, CPU clock, and memory bus cycle times. The main problem here is to avoid starving the cache. One has to balance how many operations are done by the CPU for each byte that comes from/to RAM and make sure that the timing is right. All these factors vary a lot between different CPU, mobo, and RAM models. To state that Altivec is either better or worse than SSE2 is simplistic, they are functionally identical and the relative performance between them will be determined by secondary factors.
The biggest problem in SSE2 is that the only compiler that optimizes it well is Intel's, gcc sucks when generating code for the P4, but with hand-optimized code this is irrelevant. If the Intel architecture that Apple will adopt has SSE2, this could be very good news for developers. Let's hope Apple implements efficient optimization for SSE2.
Yes, you should have read all the keynote transcripts. They did the same thing when the PowerPC came out, developers were given prototype 6100s as part of their developer kit.
Not since Marie-Antoinette played milkmaid has looking simple and honest been so fake and complicated.
As others have pointed out elsewhere in this thread, the dev kit includes the hardware. They aren't just shipping a Tiger CD you pop into your dell.
Disclaimer: I am an OpenOffice.org Mac OS X devleoper and a founder of the NeoOffice project
Quote: This means OpenOffice.org 2.0 will work *now*. This means no more second-class Mac versions of popular OS apps.
This statement couldn't actually be farther from the truth. In fact, it will actually make the push for OpenOffice.org, at least, more difficult. If you dig into the details it means there's much more work ahead:
Changing processors does nothing to help OpenOffice.org development on Mac OS X except slow it down yet again. Chances are you'll probably see it running in an emulator for a long time before it's running on Mactel hardware.
ed
Apple posted Intel Universal Binary documentation to their website. It's interesting, and everyone should read it. Notable is a caveat that OpenFirmware is going away. That seems to point towards more standard hardwware.
Quite simply, Intel no longer uses CISC. Sure the instruction set is CISC, but it's all microcode reduced to RISC instructions underneath the hood (which was done WAAAY back with the Pentium II and may have partially been implemented on the original Pentium). MMX has been dead for a while, replaced by SIMD and SIMD2, which can actually run in parallel to the floating point unit and no longer requires a context switch. Seriously, though, outside of the math world, you probably don't need either unless you're doing software rendering of graphics - the original reason for MMX was to speed up processing of games and video effects in software and this work is now pretty much entirely handled by the GPU.
According to Apple's Universal Binary whitepaper (http://developer.apple.com/documentation/MacOSX/C onceptual/universal_binary/universal_binary.pdf), the Intel Macs will NOT be using Open Firmware...
I can't wait to see what people are able to do (legitimately or not) with the x86 dev boxes...
Apple's gcc supports universal (fat) binaries. XCode uses gcc for compilation, but it's just a wrapper. Gcc still works fine without it.
These new x86 processors have way more innovation in them than the 10 year old new and clean design of the PPC. They are not CISC processors anymore in that they internally are RISC/CISC hybrids. The PPC has grown in the other direction, and is now also a RISC/CISC hybrid.
From an architectural standpoint, PPC is still a lot cleaner than x86. But the immense brainpower and $ that Intel has put behind x86 made it into something that is hard to beat even with a cleaner design.
In the end, it's a matter of priorities: Intel had to go low-power and had the resources to develop this technology while the company line was going in the exact opposite direction (P4). Now they are killing with it. Even AMD is way behind regarding low power chips.
IBM never wanted to commit the resources or people to make the G5 portable. They would have had to spend serious money - chip design is extremely expensive - and hire very very good people. IBM never had this commitment.
Besides, I have this feeling that the G5 was designed with some P4-envy in mind: Huge pipeline, high clock speeds. And using lots of power and generating lots of heat... Intel had the Pentium-M as a 'plan B' for this boneheaded strategy, whereas Apple/IBM did not.