Linked by Thom Holwerda on Mon 5th Nov 2012 23:40 UTC
Thread beginning with comment 541160
To view parent comment, click here.
To read all comments associated with this story, please click here.
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[3]: Rest assured that apple has been running arm internally
by kovacm on Tue 6th Nov 2012 21:17
in reply to "RE[2]: Rest assured that apple has been running arm internally"
I don't understand: even if Apple could get that much of a performance boost by using GPGPU on ARM, couldn't they also get the very same perf boost with GPGPU on x86?
simple answer: if they want to switch to ARM than they could use GPU to compensate speed difference between ARM and intel x86 chips!
and even on desktop x86 you will need to utilize GPU in future if you want real impressive speed gain!
at the end speed of x86 would be less and less relevant! - "it should be only fast enough for "Office-like" jobs..." and real speed (where it is needed) will come from GPU.
even AMD bet entire company on this premise! - CPU part could be anything (ARM/x86); it does not matter. GPU is one that count. look at latest AMD GPUs - they go so far that you can pass pointers from x86 directly to GPU, they add exception support, virtual memory and page tables, calls and conditional branches, integer operations with 256 GPR. it looks like complete CPU and still massive SIMD monster.
they only need "fast enough" x86 execution since real speed will come from SIMD GPU.
RE[4]: Rest assured that apple has been running arm internally
by Neolander on Wed 7th Nov 2012 06:49
in reply to "RE[3]: Rest assured that apple has been running arm internally"




Member since:
2010-03-08
anyway, it would be great if someone make deep analyze of using ARM instead of intel ivy bridge.
looking from side, all premises that it could be done are there:
1. Apple control third party developers through AppStore
2. Apple control developing tools (language, framework, compiler...)
3. Apple investing heavy in GPGPU & multiprocessing (OpenCL/GDC...)
my question is:
how much faster could professional application* run on ARM + GPGPU chips (vs intel chips) if you optimize everything (metal, compilers, frameworks)?
*FinalCut, Logic, Aperture, Adobe Suit, Cubase, Cinema 4D...?
Is Apple in position to do this:
- potentially leap frog x86 in terms of performances optimizing everything I already mention (metal, compilers, frameworks...)?
I don't understand: even if Apple could get that much of a performance boost by using GPGPU on ARM, couldn't they also get the very same perf boost with GPGPU on x86?
Unless I'm missing something, GPGPU does not sound like a major area of differentiation between x86 and ARM, since GPUs must work similarly on both architectures (there's only so many ways to compute a matrix product in silicon).
Edited 2012-11-06 18:16 UTC