Both methods will end up with architecture optimized machine code it's just a matter of changing everywhere that particular pattern was invoked, or just fixing the compiler to use different machine code for the currently used pattern. How am I wrong?
For any given "pattern", there is not necessarily a way to improve the compiler. There may be a more efficient way of writing the function, though. The non-optimization cannot universally be fixed by improving the compiler. Rewriting the code and rewriting the compiler are not two paths to the same thing. They are two
potential methods to improving speed, but any given problem is not necessarily solved equally with both.
And the OS has nothing to do with memory allocation and application separation which is usually one of the main causes of instability?
Of course not, but application stability isn't the same as platform stability. How the system addresses memory allocation relies fundamentally on how that system interacts with its hardware. Improving stability is a multilayered approach, with only the topmost layers being sufficiently abstracted to allow for platform-agnostic coding fixes.
What's the point of having a developer's release if they aren't supposed to begin development on it now?
They
are supposed to begin now. That's the point. You can only develop for Intel, so no one is wasting any effort on 10.6 PowerPC development that will be scrapped by a "delay" in the announcement.
If Apple expects developers to keep working on Universal binaries they need to be able to test the applications on both PPC and Intel. If 10.6 is Intel only Developers will have to have 10.5 systems around to test the PPC applications. How does that not make sense?
To test on PowerPC, you need a PowerPC Mac. Either way, you need two machines. If the two machines run different builds, so be it. Continue your PPC testing on the 10.5 PPC machine, and do your Intel testing on 10.6. It's not rocket science.
Have you seen the source? Your only evidence is a couple technologies that are Intel only and one is muti-core only that's nearly enough evidence to say that single core Intel is being scrapped as well.
For someone who claims to know a lot about microprocessor architecture, stating that Grand Central is evidence of "scrapping" single-core Intel is downright idiotic. Multithreaded applications don't require multiple cores, and adjusting the thread management to make it easier for applications to be multiprocessor aware doesn't change anything about their compatibility with single processor systems. Even if you didn't know this, it was made explicitly clear at WWDC, had you been there.
But now you force developers to test on multiple OS versions on multiple machines rather than a single OS version.
So? You need two machines anyway.
If I were a developer and wanted to do a speed and stability application update with no new features, knowing that I only had to work on and test one target architecture would be information I'd want to know?
You only
can work on one target architecture for 10.6.
If Adobe uses these new features in CS4 for speed and stability, but doesn't add new features, why should they milk PPC users for an update that does nothing for them,
CS4 is a new-feature release. Even if it weren't, that would simply mean that there's no reason for PPC users to update.
Every reason you give for Apple to halt PPC support is a reason for developers to want to do it as well.
Developers are free to do so whenever they like. They don't need to wait for 10.6, an announcement, or anything else. Some of them have already stopped.
They don't know what Apple is going to do unless they are privy to certain information from Apple (with non-disclosures) that they are then talking from in total violation of it.
Non-disclosure only affects those who are a party to it. Once the information is released, whether intentionally or by accident, there is no third party liability.
What kind of evidence? They JUST bought a PowerPC chip maker for goodness sake! They've said they plan to have them work on future iPhone parts....um....cough... what does that SUGGEST? No no no, PowerPC is DEAD! Get over it! Well what about ARM? Where is ARM in all of this? It's not talked about AT ALL. When you point out, hey does Snow Leopard have a developer ARM build in it? Does that mean ARM support in OSX has been dropped???? They ignore it. But that's the SAME LOGIC they're using to ASSUME PPC support internal to Snow Leopard has been dropped simply because they didn't get copy! Wow. That takes some genius level thinking to come to that conclusion.
Rant, rant, colossally poor logic, rant.
OS X and OS X iPhone are not identical. The desktop-class development seed has never and will never contain an ARM build unless some future ARM derivative becomes a desktop processor (something it's not very well suited to do). PowerPC could easily be adopted in future embedded devices and OS X iPhone could easily be developed to facilitate it.
Honestly, each time you make a claim like this, the humor value skyrockets in your admonishing of people who "don't know" what they're talking about.
That was my point entirely. It's not MIMICKING anything. It simply uses what's available. Intel just added ANOTHER 'altivec-like' function
Who said anything about mimicry, emulation, or anything of the sort?
The mere FACT Jobs already alluded that Snow Leopard will have a much smaller memory footprint means that it would be USEFUL to PPC users
Back to square one. That depends entirely on
why and
how the memory usage is reduced. Even neglecting specifically optimized hardware performance for a moment, if it is more efficient virtual memory management (likely, given that memory-consuming resources are largely non-executing modules), it will be hamstrung by PPC's poor bus performance. Any advantage would be eliminated and could even worsen performance (almost
certainly on the G4).
Bug fixes, smaller memory footprint, etc. are all HIGHLY USEFUL to PPC owners. Snow Leopard might even mean the OS could run on LOWER rated configurations compared to Leopard, bringing some 'left out in the cold' back into the fold.
Yeah, right.
Are people going to tell me that these optimizations are ONLY useful to Intel machines when their biggest product right now is a PHONE? I think I'd want Snow Leopard to work BEST on that phone since that's where every CPU cycle counts and battery power needs to be maxed, etc.
Wow. The iPhone doesn't
The know-it-alls, of course, don't believe this. They KNOW Apple has ditched PPC once and for all
There's no certainty. There are, however, several points suggesting it and exactly zero contradictory evidence.
The announcement of no new features means that preserving compatibility with Leopard isn't an issue, so any 10.6 software that is built as a UB will continue to run, just as on Leopard. Only developers who choose to drop PPC will do so--there will be nothing to force them. Writing
application software as a developer for 10.6 doesn't preclude it from running on 10.5, both Intel and PowerPC.
You don't seem to get that, after pages and pages. Apple focusing on cleaning up their Intel codebase and "laying the groundwork for the future" (Hint: more and future Intel technologies) doesn't mean that it breaks anything in 10.5. You're still looking at this like a Panther-Tiger transition, which was a software sea change. This is not that.
and anyone who believes otherwise is a whining cry-baby.
No, there are many people who believe 10.6 will end up being universal, despite all signs pointing the other way. There's nothing wrong with that. Only some of them are whining, ranting chores. A rectangle is not a square.
