Xref: utzoo comp.lang.eiffel:832 comp.lang.c++:7015 Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!image.soe.clarkson.edu!news From: cline@cheetah.ece.clarkson.edu (Marshall Cline) Newsgroups: comp.lang.eiffel,comp.lang.c++ Subject: Eiffel/C++/Smalltalk `Benchmark' is COMPLETELY INVALID Message-ID: Date: 29 Mar 90 20:37:36 GMT References: <274@eiffel.UUCP> Sender: news@sun.soe.clarkson.edu Reply-To: cline@sun.soe.clarkson.edu (Marshall Cline) Distribution: usa Organization: (I don't speak for the) ECE Dept, Clarkson Univ, Potsdam, NY Lines: 40 In-reply-to: kimr@eiffel.uucp's message of Fri Mar 23 17:36:41 1990 In article <274@eiffel.UUCP>, kimr@eiffel.UUCP (Kim Rochat) concludes: > any perception that C++ has superior performance to Eiffel may > be invalid. I enjoy solid technical articles, but unfortunately this was neither. I claim the conclusions the authors draw are Completely Invalid: (1) The algorithm is implemented DIFFERENTLY in C++ as in Eiffel. In particular, the authors ADDED FUNCTION CALLS to the C++ version! All the lowest level (highest call rate!) actions are non-inlined functions in C++, but they aren't function calls at all in Eiffel. (2) The `benchmark' problem (`mosiac') doesn't exercise ANY of the OO features of the languages. The authors even admit this, but imply it is a `feature' -- they claim it compares how the compilers handle array dereferencing -- but see the next point. (3) The bottleneck for mosiac is `random()', a common library function. Thus `mosiac' is a TERRIBLE benchmark -- it might be a valid benchmark for random() implementations, but certainly NOT for OOP. (4) The I/O was stripped, supposedly to avoid biasing results due to I/O on a workstation. But they compared Sun with Sun, so that makes no sense. (I wonder how Eiffel and C++ compare in their I/O? Anyone in netland with Eiffel/g++/Cfront want to do it?) CONCLUSIONS: Mosaic is just a big array calculation with lots of random numbers, hardly comparing the speed of Eiffel and C++ code. It's a pathetic benchmark, demonstrating Nothing about the languages in question since most of its time was spent inside a "C" library routine. Marshall Cline -- =============================================================================== Marshall Cline/ECE Department/Clarkson University/Potsdam NY 13676/315-268-3868 cline@sun.soe.clarkson.edu, bitnet: BH0W@CLUTX, uunet!clutx.clarkson.edu!bh0w Career search in progress: ECE faculty. Research oriented. Will send vitae. ===============================================================================