On Tue, Jan 4, 2011 at 2:47 AM, Miles Bader <mi...@gnu.org> wrote: >> Are you talking about the time to process those empty rules that are >> gone if you get rid of -MP or of the remaining 20 seconds? > > Both, I suppose. I imagine both have the same inefficiency (just more > in the -MP case)...
Most of the CPU time is user time, so based in your analysis the problem would be bad algos in make or whatever, or just sheer processing of such a huge file (as shown by a 23s reduction by just getting rid of empty rules). I haven't done through testing in a cold system to see how badly we are hit by I/O, mostly because my main use case is fast turnout in a warm disk. That being said I can imagine reading all the thousands of .Plo files cannot be helping, but at least in a warm system it does not seem to be too much of an issue. I did a profile with sysprof but I really didn't know enough of GNU make internals to figure out what was going on in detail; I can send the output to the list if you think it could be useful. At least for now it seems that just removing the vast amounts of duplicated info in the final makefile results in huge wins, so I'm optimistic that I can improve things without directly hacking on GNU make. Xan > > -miles > > -- > Cat is power. Cat is peace. >
