Wine 1.4 took ~8 hours to compile from source.... is that normal?

Zombie8 · #1 26th April 2012, 11:17 PM

I went out to winehq and grabbed the latest source code for the stable 1.4 release of Wine today. It took my computer roughly 8 hours to finish compiling this software. Below are my computers specs....

OS: Fedora 16
Kernel: 3.3.0-4.fc16.i686
Ram: 4gb
Processor: AMD X2 6000+

I just feel that the time it took to compile this software was somewhat outrageous. My hardware was not anything bad in my opinion so that fact that it took this long made me concerned about my machine.

Does this amount of time seem reasonable? Anyone have any ideas what may be causing the slowness on my machine? I am open to all ideas and responses.

Thanks everyone!

sea · #2 26th April 2012, 11:55 PM

Did you use the "-j #" switch to enable the # amount of thread available to the cores?
Do you have all 4 gb ram available to compile?
Just asking because you're using a i686 without PAE kernel.

However, you're right, 8 hours is a pretty long time to compile.
If the source is very hughe (several hunderds/thousends megs), this might happen, but i dont 'feel' like wine was that hughe.

Zombie8 · #3 27th April 2012, 12:03 AM

Quote:

Originally Posted by sea

Did you use the "-j #" switch to enable the # amount of thread available to the cores?

No I was unaware of this switch. I do not compile software very often so consider me a n00b when knowing any topic beyond "run configure, make, make install"

Quote:

Originally Posted by sea

Do you have all 4 gb ram available to compile?

I have 2 x 2gb ram sticks in the computer, but you are correct I am not running a PAE kernel so total available ram to the system is about 3 gbs.

Quote:

Originally Posted by sea

If the source is very hughe (several hunderds/thousends megs), this might happen, but i dont 'feel' like wine was that hughe.

You are correct, the downloaded source tarball was ~19.1 mb not very huge at all.

sea · #4 27th April 2012, 12:21 AM

As you use tags, please use [ instead of < and ] instead of >.. ty

A little script to let you know the number to use for the switch:

Code:

                tmpvars=$(lscpu|grep per)
                totalthreads=$(echo $tmpvars|awk '{print $4 * $8}')
                echo "Use \"-j $totalthreads\""

goodfella · #5 27th April 2012, 04:26 AM

Quote:

Originally Posted by Zombie8

I went out to winehq and grabbed the latest source code for the stable 1.4 release of Wine today. It took my computer roughly 8 hours to finish compiling this software. Below are my computers specs....

OS: Fedora 16
Kernel: 3.3.0-4.fc16.i686
Ram: 4gb
Processor: AMD X2 6000+

I just feel that the time it took to compile this software was somewhat outrageous. My hardware was not anything bad in my opinion so that fact that it took this long made me concerned about my machine.

Does this amount of time seem reasonable? Anyone have any ideas what may be causing the slowness on my machine? I am open to all ideas and responses.

Thanks everyone!

It does take a long time to compile Wine. If you want to try another biggy, go get
Qt4: qt-everywhere-opensource-src-4.8.0.tar.gz. I will do a compile of Wine and
give you something to compare with.

jpollard · #6 27th April 2012, 04:49 AM

One note about the -j switch.

The makefiles need to be appropriately configured for it to work really well - and sometimes odd things can happen during a recompile. If everything (all dependencies) are fully included, then it can save a LOT of time.

For instance, if a package uses 3/4 libraries, it is possible for each to be compiled in parallel, while the applications that depend on those libraries waits until they are rebuilt.

But if some of the dependencies are assumed (as in the library already exists but isn't explicitly listed as a requirement) then errors can occur due to an incomplete library.

When the kernel build system was modified several times to clean up accidental omissions when the -j option was used.

On a 686 single core machine, I wouldn't recommend more than a -j 2 (or 3). I have done 4 threads for the kernel on such, but it didn't speed things up much (5% or so) over 2 threads, and with more it actually slowed down.

goodfella · #7 27th April 2012, 03:30 PM

Quote:

Originally Posted by Zombie8

I went out to winehq and grabbed the latest source code for the stable 1.4 release of Wine today. It took my computer roughly 8 hours to finish compiling this software. Below are my computers specs....

OS: Fedora 16
Kernel: 3.3.0-4.fc16.i686
Ram: 4gb
Processor: AMD X2 6000+
!

OS: Linux Fedora 16
Kernel: 3.3.3
Ram 2gb
Processor: Intel P4 2.4GHz with no overclocking

configure message:
configure: OpenCL development files not found, OpenCL won't be supported.
configure: libhal development files not found, no legacy dynamic device support.
configure: libcapi20 development files not found, ISDN won't be supported.

configure: Finished. Do 'make' to compile Wine.

Without the above three objects on my system it took 1 hr 16 minutes to compile Wine.
This is on a system with hardware that is about 9 years young.

Zombie8 · #8 27th April 2012, 04:04 PM

Quote:

Originally Posted by goodfella

OS: Linux Fedora 16
Kernel: 3.3.3
Ram 2gb
Processor: Intel P4 2.4GHz with no overclocking

configure message:
configure: OpenCL development files not found, OpenCL won't be supported.
configure: libhal development files not found, no legacy dynamic device support.
configure: libcapi20 development files not found, ISDN won't be supported.

configure: Finished. Do 'make' to compile Wine.

Without the above three objects on my system it took 1 hr 16 minutes to compile Wine.
This is on a system with hardware that is about 9 years young.

Seeing this makes me start to believe that is something very wrong with my processor. I got it from my dad who gave up on the processor after being dissatisfied with its performance on a windows 7 system.

I tried to play minecraft the other day on my machine and it was just get laggy and the cpu would spike to 100%.

Does anyone have any ideas for how I can benchmark my processor and get some numbers to compare with? I have recently been reading articles on Phoronix and they have a test suite that I might give a try. Are there any other options out there to try as well?

Thank you very much goodfella for showing me your results and information. It was very helpful.

Can anyone else take the time to do the same for me as goodfella did? It would be greatly appreciated. Thanks all!

goodfella · #9 27th April 2012, 04:33 PM

Quote:

Originally Posted by Zombie8

Seeing this makes me start to believe that is something very wrong with my processor.

Can anyone else take the time to do the same for me as goodfella did? It would be greatly appreciated. Thanks all!

I would do this first:
make distclean then make, see if the long build time happens again.

Look at bios settings for the motherboard.

Build a 3.3.3 kernel with configuration specific to AMD processor.

If I can help you out any more just ask.

Zombie8 · #10 30th April 2012, 03:58 PM

Wanted to post an update to my situation. I had a busy weekend and only was able to gather the following information in my spare time. Keep in mind before each compilation I ran "make clean" to make sure that all previous compiled source was removed and I was starting fresh each time.

Scenario #1:
Command: time make
Time Output: REAL: 504m20s
USER: 458m7s
SYS: 33m23s

Scenario #2:
Command: time{make -j 2}
Time Output: REAL: 260m33s
USER: 394m2s
SYS: 30m45s

My next steps are to run memtest against my 1 stick of 2gb ddr2 ram and post the results. Then I will try the above scenarios again but with an amd specific compiled kernel.

As a side note, I am starting to think the motherboard or processor maybe where the issue is. It is hard to tell since I do not have any backup motherboards or processors to use at the moment.

goodfella · #11 30th April 2012, 11:18 PM

If I were you I would build gcc on my machine with only C & C++ support.
You can type "gcc -v" to see what your current gcc is configured with.

Look at this link for prerequistes:
http://gcc.gnu.org/install/prerequisites.htm

I have the configuration options for gcc-4.7.0 minus ada support.
There is no support for building ada. If you are not programming
military satellites then you won't need any Ada.

../gcc-4.7.0/configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --with-bugurl=http://bugzilla.redhat.com/bugzilla --enable-bootstrap --enable-shared --enable-threads=posix --enable-checking=release --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-gnu-unique-object --enable-linker-build-id --enable-languages=c,c++,objc,obj-c++,java,fortran,go,lto --enable-plugin --enable-java-awt=gtk --disable-dssi --with-java-home=/usr/lib/jvm/java-1.5.0-gcj-1.5.0.0/jre --enable-libgcj-multifile --enable-java-maintainer-mode --with-ecj-jar=/usr/share/java/eclipse-ecj.jar --disable-libjava-multilib --with-ppl --with-cloog --with-tune=generic --with-arch=i686 --build=i686-redhat-linux

You probably could trim this down a bit.

jpollard · #12 1st May 2012, 02:40 PM

That wouldn't alter the time though, as these are external to the gcc compiler itself (though parts are reused in those).

goodfella · #13 5th May 2012, 04:34 PM

Are you configuring the build with "--enable-win64" & "--disable-win16" ?