The code for FASTLINK 2.2 and beyond compiles fine on OSF. In version 2.1 of FASTLINK it is necessary to comment out the definition of uchar in commondefs.h. No modifications are needed to the Makefile. The DEC Alpha on which I tested things has 3 C compilers available: cc, gcc, and c89. All seem to work. I do not yet have a recommendation on which of these to use.
Unfortunately, there are some problems with the implementation of floating point on the Alpha, which may interact badly with any or all of the compilers and FASTLINK. The bad behavior you get is a floating point exception and a core dump. I recoded some lines in FASTLINK to work around one manifestation of the problem, however, Martin Farrall reports that he still has a problem using the cc compiler with -O2 level of optimization (interestingly, both -O1 and -O3 seem to work).
If you hit a floating-point exception, please perform as many of the the following experiments as you can before reporting the problem to me:
You can recompile the program with the following compiler flags added:
#SYSDEP =For example, to add the flag -ieee_compliant, change it to:
SYSDEP = -ieee_compliantAlso, it would be nice to do these experiments with each of the 3 compilers (cc, gcc, c89).
If you are willing to send me your input data and an indication of which compiler/flag combination(s) led to a floating point exception, I am willing to investigate further to see if the FASTLINK code can be rewritten in such a way as to avoid the compiler bug.
To run parallel FASTLINK, refer to README.p4 for OSF-specific notes.