This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Odd stats


I have been running and comparing different transmittals through
sedris3.1 and comparing performance with sedris3.0.

I have noticed a few things.
1) Write speed appears to be slightly slower in 3.1 than 3.0. I thought
you had claimed otherwise. The slow down is negligable (about 1.5
seconds per minute of work). 

2)The resulting transmittals are *much* smaller. Using data generator
with the options of -d100,100 -g10,10 (creates 10*10 100mx100m grids
with 32x32x2 polygons per grid) reults in a transmittal of 95MB for a
3.0 write API. The 3.1 API produces the same transmittal in only 26MB of
space.

3)The number of objects reported by syntax checker is way off though.
The 3.0 syntax checker reports 4,505,969 objects in the transmittals and
the 3.1 syntax checker reports 2,662,769 objects. I think the vast
difference is the result of a change in the counting algorithm. I have
included the run for both the 3.0 and 3.1 syntax checker in the attached
output file. I am I correct in the assumption or am I dropping objects
on the floor somehow?

4)the run times (as you can see in the output file) are vastly different
between 3.0 and 3.1. 3.0 took 18minutes and 19 seconds whereas the 3.1
app takes 2 minutes and 18 seconds.

Marcus

Attachment: output
Description: Binary data


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]