Re: Low-memory compile times

Perry Harrington (pedward@sun4.apsoft.com)
Wed, 22 Jul 1998 11:55:31 -0700 (PDT)

This looks more like a scheduler issue than anything, notice that the CPU utilization
went down and the compilation time went up, but the User and System seconds were almost
the same. It's doing the same work, it's just using less of the CPU at the time. Maybe
it's scheduling, maybe it's something else consuming the time. I've got a P100 8MB
laptop, I'll download 2.1.110 and compile it, try tweaking a few things, and going
again.

--Perry

>
> I've been "benchmarking" the various dev kernels on my 8M 386-40 by
> doing a compile of the new kernel under 2.0.3[345] and then booting into the
> new kernel and compiling the same kernel under itself. For the most part,
> the compile speed has gotten progressively better since 2.1.91. 2.1.107 and
> .108 were the best, taking only about 75% longer under 2.1 than under 2.0.
> This isn't ideal, of course, but I'm willing to live with it to get some of
> the new 2.1 features. 2.1.109 was a major step in the wrong direction. It
> chewed on the compilation for over two days, and never did finish. I aborted
> the compile when 2.1.110 came out. 2.1.110 is somewhat better, but still not
> back to 2.1.108's level.
>
> Here's "time make zImage" outputs for some of the recent kernels:
>
> 2.1.107 build under 2.0.34:
>
> 7852.19user 820.82system 4:10:05elapsed 57%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
>
> 2.1.107 build under 2.1.107:
>
> 8428.68user 1267.79system 7:28:49elapsed 36%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
>
> My 2.1.108 results got lost in the shuffle, but they were similar to
> 2.1.107.
>
> 2.1.109 build under 2.0.35:
> 8112.93user 848.99system 4:19:35elapsed 57%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
>
> 2.1.109 build under 2.1.109:
> Never finished. Working time greater than two days.
>
> 2.1.110 build under 2.0.35:
>
> 8115.44user 827.86system 4:21:57elapsed 56%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
>
> 2.1.110 build under 2.1.110:
>
> 8985.52user 1583.75system 10:57:00elapsed 26%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
>
> ---
> John Campbell
> jcampbel@lynn.ci-n.com
>
> QotD: The problem that we thought was a problem was, indeed, a problem, but
> not the problem we thought was the problem.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html
> 

-- 
Perry Harrington       Linux rules all OSes.    APSoft      ()
email: perry@apsoft.com 			Think Blue. /\


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html