• 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: Array Byte swapping with SSE
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Array Byte swapping with SSE


  • Subject: Re: Array Byte swapping with SSE
  • From: Eric Albert <email@hidden>
  • Date: Thu, 27 Jul 2006 17:55:36 -0700

The primary references for the SSE intrinsics are Intel's manuals. I'd still encourage you to ask about this on perfoptimization-dev, since there's a good chance someone on that list has had to do something similar to what you need.

-Eric

On Jul 27, 2006, at 11:38 AM, Jeremy Sagan wrote:

Hello Eric,

Thanks for the quick response. I am aware of how to swap using standard methods and I understand that this is not a compute intensive operation. On the other hand what I want to know is how to do this using SSE as a small example like this will likely bootstrap me into writing the code that I need to, that simply begins with an endian swap and then does much more, like convert to float, vector multiplication and addition, etc, etc, I have looked at using the Accelerate framework but I do not see that as a good fit.

Also I am looking for a reference to the C interface to SSE. Are the only references the Intel Pentium manuals?

Jeremy

On Jul 27, 2006, at 2:24 PM, Eric Albert wrote:

On Jul 27, 2006, at 11:15 AM, Jeremy Sagan wrote:

Does anyone have any example code for endian byte swapping an aligned array of SInt16's using SSE on intel Macs?

I think is should be the equivalent of a sequence of loads, permutes, and stores but I am not sure where to find the C interface documentation for SSE and I am very interested to see sample code like this.

Unless your array is huge, you'll usually get reasonable performance from a simple for loop:


	#include <libkern/OSByteOrder.h>

	for(int i = 0; i < length; i++) {
		array[i] = OSSwapInt16(array[i]);
	}

That's the main reason why we haven't added bulk byte swappers to the system yet. We've found that byte swapping rarely shows up as a performance bottleneck for us.

If it is a performance bottleneck for you, you might want to ask about this on the perfoptimization-dev mailing list.

-Eric


_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40outofcheese.org


This email sent to email@hidden

_______________________________________________ Do not post admin requests to the list. They will be ignored. Xcode-users mailing list (email@hidden) Help/Unsubscribe/Update your Subscription: This email sent to email@hidden
References: 
 >Array Byte swapping with SSE (From: Jeremy Sagan <email@hidden>)
 >Re: Array Byte swapping with SSE (From: Eric Albert <email@hidden>)
 >Re: Array Byte swapping with SSE (From: Jeremy Sagan <email@hidden>)

  • Prev by Date: Re: XCode, Subversion and Nib files
  • Next by Date: Re: XCode, Subversion and Nib files
  • Previous by thread: Re: Array Byte swapping with SSE
  • Next by thread: Problem with breakpoints in constructors
  • Index(es):
    • Date
    • Thread