Re: New copy of a pfNode

New Message Reply Date view Thread view Subject view Author view

Don Hatch (hatch++at++hell.engr.sgi.com)
Mon, 26 Jul 1999 16:49:22 -0700

On Jul 23, 5:42pm, Gustin Prudner wrote:
> Subject: New copy of a pfNode
>
> Hello,
>
> I want to maintain a list of pfNodes that I load from disk
> so that repeated calls for the same file will simply return memory
> instead of hitting the disk. I have been unable to find a way to make a
> new copy of a pfNode. pfNode->clone returns the same reference and
> pfNode->copy does not copy GeoSets.
> Any help?
>
>
> Gustin Prudner

This isn't an answer to your question,
but make sure you do some timing tests to determine
whether your scheme is really worth the effort,
keeping in mind IRIX's disk cache--
unused physical memory is used as a cache for the disks,
so the second and subsequent accesses to a particular
section of a file should be substantially quicker than
the first access, by a factor of maybe 8
(assuming you have enough free physical memory).
Note that this cache is shared between processes and persists
from one program run to the next.
Of course your scheme will probably be even faster,
but it may not be by much.

Don

-- 
Don Hatch  hatch++at++sgi.com  (650) 933-5150  Silicon Graphics, Inc.

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Jul 26 1999 - 16:49:28 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.