I was first wondering if there could be a memory problem, i.e. not enough system memory or possible a subroutine with large local arrays which I believe get allocated on the stack. (I have run into these problems in the past on Unix machines and was able to change the stack size with the limit command. However this doesn't appear to be present under Cygwin). So I started playing around with some of the parameters I have which control the size of various arrays. Eventually some combinations let the code run, but the parameters were then to small to use for the problem I was trying to run. I can't say exactly how much memory before and after since the are many arrays scattered around in a number of common blocks. However, I may have been originally requesting 140 or so Meg scattereed across several arrays and common blocks. I probably cut it down to70 Megs when it ran. But increasing certain parameters while decreasing others let the code run with a larger total memory load while the reverse case (decreasing certain parameters while increasing others but a smaller total memory load) the code would not run.
Next, I recompiled the code using the -mno-cygwin flag and using the original set of parameters sizes so that the 140 or so Meg of memory was going to be needed. The code ran fine in this case. However, in a previous post about 3 months ago, I reported (and someone verified my claims) that a code compiled under Cygwin with the -mno-cygwin flag (or MSYS for that matter) ran 2 to 3 times slower than the same code compiled under Cywgin without the use of the -mno-cygwin flag.
So, the ultimate question is is anyone aware of memory problems, memory size limits, limits on contiguous memory chuck size limits (dealing with Fortran here) under Cygwin. I am using Cygwin 1.5.7 with g77 3.3.1.
-- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
