Re: [lmi] [PATCH] Trivial patch to avoid building unnecessary wxWidgets parts

[Greg Chicares]

On 2015-03-06 18:19, Vadim Zeitlin wrote:
> On Fri, 06 Mar 2015 18:14:18 +0000 Greg Chicares <address@hidden> wrote:
> 
> GC> I don't seem to have recorded how long it took previously, but...
> 
>  Me neither exactly because it was too long to redo it again just to
> benchmark it. It takes less than 10 minutes now (inside my 2 CPU VM, I
> should probably give it 4 of them...) which is definitely a gain.

Please say exactly what you're doing that takes less than ten minutes.
Are you actually running the same command I gave:

/lmi/src/lmi[1]$time make wx_version=3.1.0 
wx-3.1.0-md5=3c049ed2f470d81627ea5d513a06aa15 $coefficiency -f install_wx.make 
>../log 2>&1

(which does something like 'make clean' first), and using a MinGW
gcc-3.4.5 toolchain? Does your "2 CPU VM" have two cores with that
function as four due to hyperthreading, one that functions as two
for the same reason, or two cores with hyperthreading turned off?
Are you using a cached config file?

Here's how native msw does on my machine, with 16 "CPUs":
  2 CPUs * 4 cores * 2 (hyper)threads
First I built wx and throw the result away, just to rule out
cache effects. Then I rebuilt with the old 'install_wx.make';
and then I rebuilt with the new one. This is a very recent
version of wx--the one we're using in the office now.

/lmi/src/lmi[0]$echo $coefficiency
--jobs=16
/lmi/src/lmi[0]$make wx_version=3.1.0 
wx-3.1.0-md5=479cefdfe186caa68141f11a3d8164f5 $coefficiency -f install_wx.make 
>../log 2>&1
/lmi/src/lmi[0]$time make wx_version=3.1.0 
wx-3.1.0-md5=479cefdfe186caa68141f11a3d8164f5 $coefficiency -f install_wx.make 
>../log 2>&1
make wx_version=3.1.0 wx-3.1.0-md5=479cefdfe186caa68141f11a3d8164f5 -f  >  2 \
 590.87s user 261.19s system 167% cpu 8:28.51 total
/lmi/src/lmi[0]$svn update >../log 2>&1
[that command retrieved the latest 'install_wx.make']
/lmi/src/lmi[0]$time make wx_version=3.1.0 
wx-3.1.0-md5=479cefdfe186caa68141f11a3d8164f5 $coefficiency -f install_wx.make 
>../log 2>&1
make wx_version=3.1.0 wx-3.1.0-md5=479cefdfe186caa68141f11a3d8164f5 -f  >  2 \
 528.68s user 243.44s system 187% cpu 6:50.74 total

It's about half that speed in my VM:
 541.79s user 821.04s system 165% cpu 13:41.95 total

I don't see how you can do this in less than ten minutes, with
one-quarter or one-eighth as many CPUs. And here all 16 are
pegged at 100% for more than half of the measured time; let me
send you privately a screenshot of msw "task manager" that
shows this. This is 32-bit xp, but at no point during these
runs do I have less than a gigabyte of spare RAM available.
In the VM, and also in native xp, I'm using a recent version of
cygwin, perhaps a few months old; I know of no dramatic speedup
they've achieved recently.