Was Core Image developed for the Switch?

mr skillsmr skills
Posted:
in Future Apple Hardware edited January 2014
OK, I'm not a programmer and this is all above my head, so forgive me if this sounds stupid



All the new stuff in Tiger like Core Image (and maybe core data?) where the OS is doing more stuff for you so you don't have to program it at processor level... can we now say that this was purely to make the Intel transition easier?

Comments

  • Reply 1 of 8
    powerdocpowerdoc Posts: 8,123member
    Quote:

    Originally posted by Mr Skills

    OK, I'm not a programmer and this is all above my head, so forgive me if this sounds stupid



    All the new stuff in Tiger like Core Image (and maybe core data?) where the OS is doing more stuff for you so you don't have to program it at processor level... can we now say that this was purely to make the Intel transition easier?



    Sure it is, SJ claimed that for the last 5 years, mac os X was developped for both Intel and PPC chips. That's mean that Core image who is a part of mac os X will work on intel based mac computers.
  • Reply 2 of 8
    kickahakickaha Posts: 8,760member
    Don't forget Apple's recent push for 'little' things like autovectorization into the gcc codebase. AltiVec? MMX? Who cares? Let the compiler handle it!
  • Reply 3 of 8
    mr skillsmr skills Posts: 144member
    That's not quite what I meant. Of course it will work, along with everything else. What I am asking is whether the whole purpose of Core Image (or a substantial part of it) was to ease the transition to Intel by making applications hit the processor less.
  • Reply 4 of 8
    mcqmcq Posts: 1,543member
    Quote:

    Originally posted by Mr Skills

    That's not quite what I meant. Of course it will work, along with everything else. What I am asking is whether the whole purpose of Core Image (or a substantial part of it) was to ease the transition to Intel by making applications hit the processor less.



    No - the purpose of Core Image was to shift certain graphics operations over to the GPU in order to speed them up. I don't believe it was done specifically with the intent of dealing with Intel.



    Since we're dealing with essentially the same graphics hardware that the operations are being offloaded to, I expect that Core Image will also be able to offload to the GPU on an Intel based Mac.
  • Reply 5 of 8
    carson o'geniccarson o'genic Posts: 1,278member
    A little off topic...but vaguely related to OSX technologies. I wonder if the Intel based render farm at PIXAR is or will be using OSX? I was alwasy waiting to hear that they would go with Xserves as evdience that Jobs takes his own hardware seriously. Now I think the answer is just the opposite.
  • Reply 6 of 8
    dave k.dave k. Posts: 1,306member
    Quote:

    Originally posted by Powerdoc

    Sure it is, SJ claimed that for the last 5 years, mac os X was developped for both Intel and PPC chips.



    So Apple maintains double the programming staff of Mac OS X? One set of programmers for PPC and another for Intel?
  • Reply 7 of 8
    macserverxmacserverx Posts: 217member
    Maybe, prolly not double; but Apple hires the best of the best of the best programmers and engineers and nerds and geeks. The guys writing code for PPC versions are likely smart enough to make things good for cross-compiling, with a special team that takes the source changes from the day before and works at them to get them to build and submits their own changes back.



    This is probably the ultimate reason that Apple is expanding their OS release timetable, keeping track of all the modified code is becoming very tedious.



    In relation to this topic, the fact that CoreImage compiles CoreImage kernels on the fly is in part due to architecture as well as simply the GPU variability.



    I'd also bet that the Pixar Renderfarm is running Linux (something from Sun even), that has absolute minimal overhead.
  • Reply 8 of 8
    kickahakickaha Posts: 8,760member
    Quote:

    Originally posted by Dave K.

    So Apple maintains double the programming staff of Mac OS X? One set of programmers for PPC and another for Intel?



    Not even close.



    The whole point of programming languages is to make it so you don't have to write to a specific chip. You literally just tell the compiler which chip you want to run on, and it does the right thing. gcc is very good at this. Xcode 2.1, from the screenshots, literally makes this a checkbox option.



    Ensuring that it could run on another architecture at all times is also just good engineering sense - you want to make sure that you haven't run yourself down a rabbit hole with assumptions about one particular architecture.



    gcc + portable code means Apple can pick and choose CPU vendors, families, and designs, based on the particular product it wants to put into the market, and not have to worry about supporting a plethora of codebases, ala MS.



    It also means that Apple can play the chipmakers off of one another much more easily. Heck, if IBM comes roaring out of the gate with a kick-ass server chip on a PowerPC core, Apple is just as ready for it as if Intel comes out with an unbelievably fast low-power chip for a tablet. They can mix and match, and move with the market like no one else has been able to yet.



    That being said, it's not going to be all wine and roses. (Or is that WINE and Rosettas? Ah, I slay me.) It's going to be bumpy for some folks, and a relative cakewalk for others. The *worst* task ahead is the marketing. The technology end doesn't worry me in the least.
Sign In or Register to comment.