Manta Interactive Ray Tracer Development Mailing List

Text archives Help


Re: [MANTA] memalign


Chronological Thread 
  • From: James Bigler <bigler@sci.utah.edu>
  • Cc: MANTA <manta@sci.utah.edu>
  • Subject: Re: [MANTA] memalign
  • Date: Sun, 08 Apr 2007 21:55:07 -0600

Your solution works fine if the alignment is 16 or less. It won't work for the general solution (like posix_memalign does).

It would be nice if there was some kind of assert that could make sure if someone did want something more it blows up here instead of somewhere else.

I'll go ahead and add one.

Also, I was trying to figure out a way that would allow the deallocation of a pointer that was moved around (like memalign or self coded one does). This would require some book keeping, though. posix_memalign pointers can be freed directly. I guess I can cook something up after the site visit.

James

Steven G. Parker wrote:
Thanks, solomon!
Steve

On Apr 6, 2007, at 1:04 PM, Solomon Boulos wrote:

Oh yeah, I forgot to commit this change. In any case, luckily on Mac OS X you don't need it (committed now)

On Apr 6, 2007, at 1:02 PM, Steven G. Parker wrote:

James, bad news - memalign is not available on OS/X...

Steve








Archive powered by MHonArc 2.6.16.

Top of page