Hi, you are right :) I am happy to see both cpu/cores doing the compiling :)
greetings alex Am Freitag, den 13.11.2009, 18:43 +0000 schrieb David W Noon: > On Fri, 13 Nov 2009 17:38:20 +0100, Alexander Kaupp wrote about > [Lazarus] make option -j: > > > I dont know if everyone knows this: > > > > I have a dualcore CPU now and was wondering how to use both cores for > > compiling for example lazarus. > > > > make -j2 all > > > > uses both cpus and works great with lazarus :) > > > > for a quad-core use option: -j4 > > > > hope someone finds this useful > > This is not wholly dependent on having an SMP hardware configuration. > > Even on a uniprocessor system, make can be accelerated by specifying > -j2 or -j3 or even higher. This is because it is not the make program > exploiting multiple processes but the Linux (or other) kernel having > multiple processes to despatch (or "schedule", as UNIX boffins misname > this). It doesn't need multiple CPUs, just time-slicing and enough RAM. > > Of course, the more processor cores you have on your system, the higher > the number that can sensibly follow -j. The rule-of-thumb I use for > make is: > > 1. Let n denote the number of CPUs installed on the system. > > 2. Try 3*n+2 for the number to follow -j. > > 3. If that number seems too high, typically by causing paging, then try > 2*n+1 instead. > > So, for a monadic system, I try -j5 first, then -j3 if the machine > pages. > > For a dyadic system, I try -j8 first, then -j5 if the machine pages. > > For a tetradic system, I try -j14 first, then -j9 if the machine pages. > > ... And so on. > > But that is not a hard and fast rule, just a rule-of-thumb. > > Note: > ===== > Monadic = 1 CPU > Dyadic = 2 CPUs > Tetradic = 4 CPUs > > These terms came into use in the 1960s when mainframes first acquired > SMP (or "polyadic") configurations. > > My dream is to have a hexadekadic box under my desk!! You should now be > able to work out what that means, from the above. ... ;-) -- _______________________________________________ Lazarus mailing list [email protected] http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
