• 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: altivec problems
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: altivec problems


  • Subject: Re: altivec problems
  • From: Stephen Davis <email@hidden>
  • Date: Mon, 8 Nov 2004 11:22:08 -0800

On Nov 8, 2004, at 10:50 AM, Ian Ollmann wrote:

On Nov 8, 2004, at 10:41 AM, alex wrote:

This is not true.

The VRSAVE functionality is a compiler switch. The reason behind this is that some developers need to control the VRSAVE register directly and don't want the compiler to generate this code. There are also cases where you are calling nested functions that don't require using VRSAVE as an optimization in tight loops, etc.


Alex,

Can you be more explicit? I am not familiar with a compiler switch to turn vrsave off. By default the compiler should and does generate vrsave code (except when misbehaving).

Ian

The CodeWarrior compiler has #pragmas for turning on/off the automatic VRSAVE generation for the reasons Alex mentions.


As for the original question, if os is only using inline assembly, *maybe* gcc lets you sneak through altivec code even if "-faltivec" is not specified. The gcc in-line assembly is too grotesque for me to just try this theory. :-) os, what version of the tools are you using to build your code?

stephen

_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden


References: 
 >Re: altivec problems (From: Ian Ollmann <email@hidden>)
 >Re: altivec problems (From: alex <email@hidden>)
 >Re: altivec problems (From: Ian Ollmann <email@hidden>)

  • Prev by Date: Re: altivec problems
  • Next by Date: Re: altivec problems
  • Previous by thread: Re: altivec problems
  • Next by thread: Re: altivec problems
  • Index(es):
    • Date
    • Thread