Need to improve COM performance

[Tom Fenn]

Hello,

I'm working on an application that involves calculating some
statistics for data produced by a commercial scientific instrument
(ion trap mass spectrometer).  The code works, but takes longer than I
expected to execute.  I ran the profiler on it, and got the following
results (which are sorted on cumtime and percall).  Access to the data
file is through a COM control provided by the manufacturer.  I've
wrapped it using genpy, and then wrapped an easier to use class around
the component.  If I read the profile data right, the program is
spending 14 of its 15 minutes of execution time in the ApplyTypes
routine from PythonCOM. Is there anything I can do to speed this up
from within Python?  Would implementing the Python class in C++
improve performance significantly in this case?  Happily, the
application is spending only a few seconds in my own code.

ncalls	tottime	percall	cumtime	percall	filename:lineno(function)
0	0	0	profile:0(profiler)		
1	0.245	0.245	893.4	893.4	profile:0(r
1	0	0	893.155	893.155	<string>:1(?)
1	1.103	1.103	893.155	893.155	coda.py:4(score)
926	2.159	0.002	888.736	0.96	xcalibur.py:240(XIC)
3704	879.47	0.237	882.8	0.238	__init__.py:342(_ApplyTypes_)
926	0.094	0	875.341	0.945	2DDE1FB3-A60F-11D2-BD82-0060978EEB9Cx0x1x0.py:1412(GetChromatogram)
2778	0.295	0	7.848	0.003	__init__.py:345(__getattr__)
3704	0.539	0	3.854	0.001	__init__.py:91(Dispatch)
3704	0.714	0	3.33	0.001	__init__.py:365(_get_good_object_)
926	1.08	0.001	3.285	0.004	coda.py:19(mcq)
12038	0.571	0	1.921	0	__init__.py:359(_get_good_single_object_)
3704	0.863	0	1.762	0	__init__.py:17(__WrapDispatch)
1852	0.2	0	1.589	0.001	stats.py:231(__call__)
3704	0.327	0	1.553	0	dynamic.py:75(_GetGoodDispatchAndUserName)
3704	1.226	0       1.226	0	dynamic.py:67(_GetGoodDispatch)
926	0.132	0	1.163	0.001	stats.py:2730(astdev)
9260	0.695	0	1.061	0	__init__.py:369(<lambda>)
926	0.281	0	1.031	0.001	stats.py:2705(avar)
3704	0.443	0	0.718	0	gencache.py:117(GetClassForCLSID)
1852	0.361	0	0.548	0	stats.py:2040(amean)
926	0.293	0	0.369	0	Numeric.py:181(convolve)
926	0.196	0	0.366	0	stats.py:4835(ass)
2778	0.36	0	0.36	0	__init__.py:351(__setattr__)
926	0.328	0	0.328	0	xcalibur.py:56(__init__)
1852	0.293	0	0.293	0	stats.py:4774(asum)
1852	0.247	0	0.247	0	Numeric.py:565(sum)
3704	0.182	0	0.182	0	__init__.py:310(__init__)
3704	0.139	0	0.139	0	CLSIDToClass.py:52(HasClass)
3704	0.135	0	0.135	0	CLSIDToClass.py:45(GetClass)
1852	0.126	0	0.126	0	Numeric.py:514(ravel)
1852	0.09	0	0.09	0	__init__.py:129(__getattr__)
926	0.076	0	0.076	0	Numeric.py:122(asarray)
926	0.073	0	0.073	0	2DDE1FB3-A60F-11D2-BD82-0060978EEB9Cx0x1x0.py:4501(IsTraceValid)
930	0.033	0	0.033	0	UserDict.py:14(__getitem__)
926	0.03	0	0.03	0	xcalibur.py:81(AsArray)