There is always the possibility of paging RAM to disk at any point for any bit of memory and this goes for something like pix_buffer or Pd's table/array too. Obviously, the machine needs to have not only the free memory for the RAM disk but a comfortable amount of headroom.
The OSX memory manager won't always do the right thing, but one can try to minimize its stupidity. A lot of the kernel code is still based on assumptions about ancient hardware and non-existent server scenarios, and it certainly pays the price for it. On 7/30/07, Max Neupert <[EMAIL PROTECTED]> wrote: > > hi, > some sources suggest that there is no or little advantage to use a > ramdisk in os x because the system caches data which has frequent > access anyway. > so a ramdisk can be counterproductive in a cenario where you have > little ram: > lets say you have one gig memory. 300 mb are used by some > applications which are open, and another 300 by the system. now you > have a video in your ramdisk of also 300. the system tries to cache > it because it wants to accellerate but can't allocate enough anymore > and data will be swaped on the harddisk.. > so you end up with less performance and more harddrive access because > of the ramdisk.. > > i'm curious about your findings. > > greetings, m. > > > http://lowendmac.com/x-basics/02/0201.html > http://www.kernelthread.com/forums/viewtopic.php?t=138 > > >
_______________________________________________ [email protected] mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
