Home »
Public Forums »
archive »
Re: Shared Memory in Windows
| Re: Shared Memory in Windows [message #48224] |
Wed, 05 April 2006 19:11  |
Robbie
Messages: 165
Registered: February 2006
|
Senior Member |
|
|
Thanks for the heads up on the Windows page file. I ran some test code
to observe it using pagefile.sys when the segment was larger than RAM.
It's very comforting to see the theory match the practical results.
> The first rule of performance tuning is to only tune when
> you have to. Try it out -- you may find that the pagefile works fast
> enough for your needs.
Well, I'm actually investigating shared memory so I can make a decision
on my data model. I'm mid-way through my project and I have
deliberately left the data model quite open. However, I can't leave it
open forever or else I will start to loose sleep.
I'm interested in using various C/C++ libraries such as ITK to perform
image filtering and registration. A significant incompaibiltiy is that
ITK uses data pipelines rather than committing pixeldata to memory. My
own IDL code actually implements a poor mans version of data pipelines,
with a mid-sized granularity (currently frame by frame because it works
well with IDLgrWindow::QueryRequiredTiles). After writing a very
butchered implementation of pipelines in IDL, I can see this project
spinning out of control when I try to write C wrappers for the ITK data
model. I also have to realize that ITK isn't the only imaging library
which I might want to use.
Shared memory allows me to be a bit more generic and it makes design
and testing much easier. The downside is that the space used by all
opened images cannot exceed the size of RAM (At least with POSIX
anyway). This is problematic if the UI keeps a track of intermediate
images.
Robbie
|
|
|
|
Current Time: Wed Oct 08 15:37:51 PDT 2025
Total time taken to generate the page: 0.00388 seconds
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
