Re: NetBSD/vax 11.0RC1
"Maciej W. Rozycki" <[email protected]>
| Newsgroups | gmane.os.netbsd.ports.vax |
|---|---|
| Message-ID | <[email protected]> |
On Mon, 16 Feb 2026, Paul Koning wrote: > > I question the use of `-j' with an emulator unless it can fork/clone into > > multiple processes/threads on a multiprocessor, however with real hardware > > it does make a lot of sense to say `make -j <ncpu>' to utilise the extra > > CPUs which are otherwise busy with twiddling their thumbs, so if you have > > a multiprocessor VAX, say 4-way, then `make -j 4' should give you benefit. > > That does hit Makefile quirks sometimes though, so watch out for issues! > > The other reason for -j, even on a uniprocessor, is if things spend a > fair amount of time waiting for I/O completion. Compilation is usually CPU-bound though. Unless it has to wait for VM to be paged in/out, in which case better avoid `-j' anyway. Parallelising according to the number of available hardware execution threads seems a reasonable rule of thumb. Maciej