Hi!
I was reading the online article called "Performance Comparison of Data Structure Implementations in Fortran 90" available at
http://www.cs.rpi.edu/~nortonc/OOF90/penalty.htmland decide to run the comparison source code with IFC 9.1.025, adding one more feature to that code, namely the new allocatable structure components feature. Speaking of performance, the new feature (allocatable components) was leading the competition with 'ADT method' (pointer components) just behind until I turned on the /check:bounds compiler option. What surprise me most was the huge performance hit (factor>10) caused by this compiler option on allocatable and pointer array components against the static allocation method or the named 'layer method'. Is that larger performance hit (of /check:bounds option) normal when using allocatable or pointer components? Why? Additionally, such a big performance hit seems not to happen with ordinary (not components) allocatable or pointer arrays. Thanks for your comments,
Bernhard Enders.
p.s: to whom it may concern, the comparison test code follows attached.
