[m-rev.] diff: .NET back-end fixes
Fergus Henderson
fjh at cs.mu.OZ.AU
Tue Feb 18 11:04:05 AEDT 2003
On 17-Feb-2003, Peter Ross <pro at missioncriticalit.com> wrote:
> On Sun, Feb 16, 2003 at 10:36:46AM +1100, Fergus Henderson wrote:
> > Estimated hours taken: 3
> > Branches: main
> >
> > Various fixes for the .NET back-end.
> >
> > compiler/Mmakefile:
> > For IL grades, use editbin to set the stack size to 100M,
> > since the default (1M) is too small.
>
> A possible problem with this is that you can exhaust VM because each
> instance of the compiler will reserve 100M in the VM system on windows
> systems.
Are you sure this is system-wide rather than per-process?
There's settings for both "reserve" and "commit". I've only set the
"reserve" value, not the "commit" value, so this should only reserve
address space. Unless Windows is a much crappier OS than I thought, there
should be 4G address space *per process*. A big chunk of this will be
reserved for the operating system, but there should still be at very
least 1G left for the user process, and I think more likely 3G.
Allocating 100M of the address space for the stack still leaves at least
900M for code, static data, and heap space, which should be plenty.
If I had set the commit size, then yes, that might be a problem,
since then the OS would need to reserve 100M of swap space for
each invocation of mercury_compile.exe. But I didn't set the
commit size, so AFAIK it shouldn't be a problem -- right?
--
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-reviews mailing list
post: mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------
More information about the reviews
mailing list