|
Boost : |
From: Adrien Gervaise (adrien.gervaise_at_[hidden])
Date: 2005-08-19 03:51:55
On Thu, 18 Aug 2005 19:05:13 +0200, Robert Ramey <ramey_at_[hidden]> wrote:
> One question - "could reduce..." how much time did it actually save?
Here are the results of some tests we performed when investigating the
problem :
The data set can be considered a random graph of objects. It is about 30MB
large and is serialized several times in a row.
Save number Without the pool With the pool
1 00:00:12 00:00:12
2 00:00:31 00:00:12
3 00:00:54 00:00:13
4 00:00:56 00:00:13
5 00:00:58 00:00:13
6 00:01:03 00:00:13
7 00:01:03 00:00:13
8 00:01:02 00:00:13
On the real system, the size of the saved data can go up to a 100MB.
On a side note, the stack size of the application had to be increased to
15MB in order to avoid stack overflows during the serialization process.
-- Adrien Gervaise MasaGroup SCI
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk