Native RMI optimization

Tue Mar 21 13:16:00 GMT 2006

Hi all, 
Trying to make GCJ-compiled native RMI to run faster than Java RMI for
the last couple of days, but the result is
rather discouraging. Would somebody help me out here? 
My understanding of GCJ-compiled native RMI is that it should run faster
than Java RMI. Having said that, I did a test to see the speed of pure
RMI calls. By pure, I mean client calls a empty remote method. If I
measure the time in between, this should give me the RMI call time,
right? Here is a set of results I got (I have also attached my source
code and shell script with this email): 
Numofcalls	 Java RMI(ms)	Native RMI(ms)
300.0 	 142		172 
400.0 	 235		77 
500.0 	 338		112 
1000.0 	 661		189 
2000.0 	 1111		530 
3000.0 	 1592		908 
4000.0 	 1782 	1216 
5000.0 	 1776 	1508 
10000.0 	 3083 	2933 
15000.0 	 4268 	4404 
20000.0 	 5309 	5874 
25000.0 	 6603 	7386 
30000.0 	 7311 	8775 
35000.0 	 9277		10343 
40000.0 	 10638 	11751 
45000.0 	 11575 	13490 
50000.0 	 13362 	14652 
55000.0 	 14692 	16106 
100000.0 26454 	29374 
Oddly enough, when the number of calls are small, the native rmi is
faster than java rmi, but when the number grows bigger, it actually
becomes slower. Any ideas why? I know GCJ version of RMI has not been a
focus of the development, would this be the reason? Or is it something
to do with my code? 
In my implementation, O3 is used for the compile-time optimization (see
my shell script), and there is no thread or huge amount of objects in
the code. ( the time is measured using Sun hotspot 1.5 and GCJ 4.2)
Thank you very much
Rui 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nativermi.rar
Type: application/octet-stream
Size: 27529 bytes
Desc: nativermi.rar
URL: <http://gcc.gnu.org/pipermail/java/attachments/20060321/1cf2d3f3/attachment.obj>