Could you run those same tests on a local disk? It would be fun to see if it's only you Sun guys who have NFS who are taking the hit .. :-)
Kai On 22 Jan 2007 19:33:33 +0100, Thorsten Behrens <[EMAIL PROTECTED]> wrote:
Hi *, so - finally got around doing some build-time benchmarking. Setup: NFS volume on Solaris10, 4NT build on WinXP SP2, Opteron 2.2GHz machine, 100MBps NIC, msvc 7.1. Both src & solver on the remote volume, as well as compiler. This should be about the worst-case scenario we've got here in Hamburg. Plain out-of-the-box m198 build, from scratch, without CVS checkout, but _with_ all Sun internal source: 37.5 hrs Patched m198 build (kendy's makedepend improvements, applied strip-guards script + some fixes for formely unreachable, now build-breaking include statements), again with all Sun internal sources: 50.5 hrs That's about 35%, or one third, increase in build time. Well. Looks like we should wait for MSVC 8, before dumping all external guards - what should be safe, right now, would clearly be to gradually pass out external header guards, e.g. by removing them en passant during bug fixing, or not introducing new guards, when adding include statements. Cheers, -- Thorsten --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
-- Kai Backman, Software Engineer, [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
