FYI
-----Original Message-----
From: Norman Vine [mailto:[EMAIL PROTECTED]]
Sent: Thursday, April 11, 2002 11:41 AM
To: '[EMAIL PROTECTED]'
Cc: '[EMAIL PROTECTED]'; '[EMAIL PROTECTED]'
Subject: RE: More Tweaks
Tony Peden
>
>--- Norman Vine <[EMAIL PROTECTED]> wrote:
>> Jon, Tony
>>
>> FYI
>> profiling shows that the blasted property module is
>> a time sink unless you hardwire in all the nodes.
>
>Hmm. What do you mean by hardwire?
CACHE ALL THE NODES YOU ARE EVER GOING TO USE
ie 'hardwire' a 'C" pointer directly to the node position in the property
tree so you only go through the fgGetNode("") Interface once as
fgGetNode() is a KILLER performance wise !!!!
The profiling statistics are quite 'clear' about what is going on
< these are 'ordinal' cost rankings for FGFS function calls >
rank before after 'caching nodes'
copy_to_JSBsim 31 46 :-)
copy_from_JSBsim 18 132 :-)))) !!!
gprof results
BEFORE MY TWEAK
0.00 0.31 167824/167824 fgUpdateTimeDepCalcs(void)
[10]
[15] 6.0 0.00 0.31 167824 FGJSBsim::update(int) [15]
0.00 0.14 167824/167825
FGJSBsim::copy_from_JSBsim(void) [18]
0.00 0.12 167824/167938 FGFDMExec::Run(void) [21]
0.00 0.05 167824/167825
FGJSBsim::copy_to_JSBsim(void) [31]
0.00 0.00 1/1 FGJSBsim::do_trim(void)
[73]
AFTER HARDWIRING the property nodes
0.00 0.10 218146/218146 fgUpdateTimeDepCalcs(void)
[3]
[15] 3.0 0.00 0.10 218146 FGJSBsim::update(int) [15]
0.00 0.09 218146/218278 FGFDMExec::Run(void) [16]
0.00 0.00 218146/218147
FGJSBsim::copy_to_JSBsim(void) [46]
0.00 0.00 1/1 FGJSBsim::do_trim(void)
[65]
0.00 0.00 218146/218147
FGJSBsim::copy_from_JSBsim(void) [132]
100% Speedup
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
