Nikos Chantziaras wrote: > On 08/03/2009 03:11 PM, Walter Dnes wrote: >> >> Actually, I put -j1 into my make.conf after being bitten by -j2 a few >> times. It doesn't slow down the emerge that much, and doesn't slow down >> the compiled program at all. And the big thing is that it has probably >> saved me from wasting time trying to track down weird emerge failures. > > Well, that's you. -j2 needs only 1/2 of the time to emerge things on > this machine, and -j4 only 1/4 on a 4 core machine. > > So my advice for others is to not take Walter's advice and use -j1 > because "it doesn't slow down the emerge that much". It can slow it > down. Up to four times slower. > > It'll actually be somewhere inbetween the two of you. None of the steps in an emerge *except* the compilation stage will be affected by the -J option. That compile stage though will be affected as you said. For some packages, that will mean that the J option has almost no effect (becuase they have multiple ./configure runs that take far longer than the compile part anyway, or because they are so small the install stage takes longer or whatever).
For most packages it will show an improvement, but almost never x2 or x4.... For me, a kernel "make" with the -j=2 option is a lot quicker than without....
signature.asc
Description: OpenPGP digital signature