This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Memory for large arrays in cygwin/g77

Dave Korn wrote:
From: cygwin-owner On Behalf Of Mark Hadfield

Dante R. Chialvo wrote:

2) Use g77 using the Wl option, ie,

g77 -O2 -o mybigprogram -Wl,--stack,100000000 mybigprogram.f

That's odd. Increasing the stack size definitely does not work for me. It just causes the program to terminate silently, with no output.

It's more random than that. The size of the stack isn't the critical issue here, since the array isn't on the stack, it's in the .bss section. The issue is the knock-on effects that specifying different initial stack sizes have at the time the executable image is mapped into memory, and the subsequent allocation or availability of the memory area for the cygheap.

Woooaaa (not "whoa" meaning "stop" but "woooaaa" as in the word used ca 1995 in movies like "Wayne's World" to indicate a combination of surprise and puzzlement).

So, Dave, we've established that Cygwin g77 is not reliable for large data arrays. In an earlier message in this thread you cited a couple of messages that gave some gory details:

I must say they're a bit too gory for me to understand.

Is there a cure that would allow a simple-minded, grey-haired Fortran programmer like me to rely on Cygwin g77 (or gfortran) for moderate-sized computational tasks?

Mark Hadfield            "Ka puwaha te tai nei, Hoea tatou"
National Institute for Water and Atmospheric Research (NIWA)

-- Unsubscribe info: Problem reports: Documentation: FAQ:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]