[mercury-users] intermodule optimisation very flaky
Fergus Henderson
fjh at cs.mu.OZ.AU
Mon Jan 27 13:05:24 AEDT 2003
On 27-Jan-2003, Michael Day <mikeday at yeslogic.com> wrote:
>
> These days, intermodule optimisation seems extremely flaky, often
> producing errors about missing declarations in .opt files for code that
> does not appear incorrect and can sometimes be "fixed" with some arbitrary
> permutation. It didn't used to be like that, as far as I can tell.
Well, we test intermodule optimization quite carefully on a nightly
basis. We bootstrap the compiler with intermodule optimization
enabled, and we also run the complete Mercury test suite in a wide
variety of grades (asm_fast, asm_fast.gc, hlc.gc, asm_fast.gc.prof,
asm_fast.prof, asm_fast.gc.memprof, asm_fast.gc.profdeep, asm_fast.gc.tr,
asm_fast.gc.tr.debug, and hlc.par.gc) with intermodule optimization
enabled. In total, we run well over 7500 tests of intermodule optimization,
every night. These tests show no regressions.
But if you have noticed a problem, by all means please submit a bug report.
> This is not a bug report, just a general gripe, as there are programs
> where intermodule optimisation appears to make a large difference to the
> running time, and not being able to use it is extremely inconvenient. I'll
> try to submit some bug reports, but these errors seem to have a habit of
> being very difficult to pare down to simple test cases.
It's OK to submit large and complicated test cases.
Cheers,
Fergus.
--
Fergus Henderson <fjh at cs.mu.oz.au> | "I have always known that the pursuit
The University of Melbourne | of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh> | -- the last words of T. S. Garp.
--------------------------------------------------------------------------
mercury-users mailing list
post: mercury-users at cs.mu.oz.au
administrative address: owner-mercury-users at cs.mu.oz.au
unsubscribe: Address: mercury-users-request at cs.mu.oz.au Message: unsubscribe
subscribe: Address: mercury-users-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------
More information about the users
mailing list