sieve-dev ISieve and memory usage

Shi Jin jinzishuai at gmail.com
Sun Jul 20 23:18:50 CDT 2008


Hi,

I just tried the new optimized mesh in my code. Basically, I managed to
build a very simplified code against the optimized sieve library. I used the
same
ALE::PCICE::Builder::readMesh()
function to load the mesh form files.
Since this one is statically sized, I assume it would take less memory than
the old one.
However, I just did a simple test. For a grid with 7680 elements, 16128
faces, 10148 edges and 1701 vertexes, my old code needs 44.38MB of memory
while the new code takes 51.61MB.
I should mention that this test is for the grid only, ie, no physical field
were constructed on top of it. Still, I am very surprised by the memory
performance. Is there a explanation to it?
I am very cautious here since my code is working but I have to reduce its
memory usage in order to run the problem sizes I have in mind.
Thank you very much.
Shi

-- 
Sincerely,
Shi Jin, Ph.D.
http://www.ualberta.ca/~sjin1/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mcs.anl.gov/mailman/private/sieve-dev/attachments/20080720/50521722/attachment.htm>


More information about the sieve-dev mailing list