• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: Mach-O Speed ?!?!?!?!?
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Mach-O Speed ?!?!?!?!?


  • Subject: Re: Mach-O Speed ?!?!?!?!?
  • From: Chilton Webb <email@hidden>
  • Date: Mon, 4 Feb 2002 22:02:35 -0600

I've been going on the exact opposite premise, that the goal of all great and powerful developers should be to move their Carbon apps to Mach-O as soon as possible in order to make their software run faster. Was I wrong?

-Chilton

On Monday, February 4, 2002, at 07:02 PM, Matt Ronge wrote:
I was browsing along Cocoa websites and docs when I came apon this:

"MachO is the binary format and ABI used natively in MacOS X. Although it
works, it's somewhat slower than, for example, CFM. The reason is that it
was designed for CISC processors (m68k, x86), and does not follows the
de-facto RISC ABI declared by Motorola & IBM. For this reason, it uses wierd
outdated PC-relative addressing of data, completely leaves register r2
(RTOC) ignored, and much more oddities that are not present in other PowerPC
ABIs. The usage of MachO saved Apple about 6 months of changing the
development tool chain to support more modern ABIs, but at the price of
30-40% performance loss (according to tests made by some kernel/lowlevel
AppleComputer engineers I know). --SlavaKarpenko

I can't believe Apple is not addressing this. A 30-40% performance boost for
OS X would make the difference in usability, IMO"


References: 
 >Mach-O Speed (From: Matt Ronge <email@hidden>)

  • Prev by Date: Re: Microseconds and Carbon
  • Next by Date: Re: Speed of Quartz (was: optimizing compilers)
  • Previous by thread: Mach-O Speed
  • Next by thread: Microseconds and Carbon
  • Index(es):
    • Date
    • Thread