This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC 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: make -j lots building and testing gcc


Bill Northcott <w.northcott@unsw.edu.au> writes:

> When running the test suite with 'make -k -j 20 check'  there is never
> more than one instance of cc1 or cc1plus running although there may be
> up to 4 instances of expect.

DejaGNU does not parallelize, and neither does the custom script used
for the Ada test suite.  Thus, the maximum number of concurrent
processes under 'make -k -jN check' is the number of distinct
testsuites, which I believe is approximately 10 with all front ends
enabled.

You can in theory get up to that number of concurrent compiler
processes, not just expect processes.  I suspect what you saw is a
sampling artifact - most of the test cases are quite small, and so the
compiler will complete before top refreshes.

As Andrew pointed out upthread, 20001226-1.c after macro expansion is
a very large piece of code.  It is relatively normal for it to cause
the compiler to run out of memory.  I suspect that most of your
failures are down to something else.  (And yes, moving the build
directory breaks stuff.)

zw


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