Eric Christopher <echristo at gmail.com> writes:> Nope. If you want it then sooner is always better than later.Ok.> Also keep in mind that some warnings are only enabled with > optimization when building with gcc.Yep. I have a quad-core machine. Can I put two buildslaves on it, each doing a -j2 build? -David
On Thu, Jan 3, 2013 at 10:55 AM, <dag at cray.com> wrote:> Eric Christopher <echristo at gmail.com> writes: > > > Nope. If you want it then sooner is always better than later. > > Ok. > > > Also keep in mind that some warnings are only enabled with > > optimization when building with gcc. > > Yep. I have a quad-core machine. Can I put two buildslaves on it, each > doing a -j2 build? > >Sure, it may just be slow. -eric -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20130103/65ec23bd/attachment.html>
On Thu, Jan 3, 2013 at 10:55 AM, <dag at cray.com> wrote:> Eric Christopher <echristo at gmail.com> writes: > >> Nope. If you want it then sooner is always better than later. > > Ok. > >> Also keep in mind that some warnings are only enabled with >> optimization when building with gcc. > > Yep. I have a quad-core machine. Can I put two buildslaves on it, each > doing a -j2 build?Essentially, yes. (you install the buildslave on the machine, you create two builder configurations in the buildmaster and assign the slave to both those builders. The jobs parameter to the builder specifies how much concurrency (so 2)) Not terribly important, but I'd probably set it up the other way - assign both builders to the machine but configure it such that only builder is using it at a time but with full (4, in your case) concurrency. That way you get some results sooner, rather than both results later.
David Blaikie <dblaikie at gmail.com> writes:> Not terribly important, but I'd probably set it up the other way - > assign both builders to the machine but configure it such that only > builder is using it at a time but with full (4, in your case) > concurrency. That way you get some results sooner, rather than both > results later.Makes sense. Thanks! -David