[aspectc-user] A General Question about Compilation Time

[Sergio Queiroz]

Hi, Olaf!

When I try to compile an example with ac++ 0.92 and ac++ 1.0, using SystemC,
there is a huge difference in the compilation time.

To compile the same files, it takes more than 7 minutes with ac++ 1.0 and less
than 1 minute with ac++ 0.92, so it is almost prohibitive to me to use ac++
1.0 because this long compilation time.

Do you have any idea about the reason of this big difference?

I am sending the files and the logs.

[]s,
Sérgio


---------- Original Message -----------
From: Olaf Spinczyk <Olaf.Spinczyk at informatik.uni-erlangen.de>
To: Sergio Queiroz <sergio at consiste.dimap.ufrn.br>
Cc: aspectc-user at aspectc.org
Sent: Fri, 18 Nov 2005 11:22:22 +0100
Subject: Re: [aspectc-user] A General Question about Compilation Time

> Hi,
> 
> ac++ 1.0pre1 shouldn't be slower than 0.9.3. According to my 
> measurements it should even be a few percent faster. 1.0pre2 will be 
> about 20% faster than 1.0pre1. However, this is still slow and we 
> will continue to improve the performance step by step.
> 
> If you want to reduce the compilation time, you could try to reduce 
> the number of translation units, e.g. like this:
> 
> project.cc:
> #include "A.cc"
> #include "B.cc"
> #include "C.cc"
> 
> ag++ -c project.cc -o project
> 
> Of course, this is not always possible.
> 
> If you know what you are doing, i.e. where your advice affects the 
> component code, you could also weave only the relevant translation 
> units instead of all files. For example, in the Puma project we 
> weave only a subset of the translation units and create transformed 
> header files. The translation units that are not woven include the 
> transformed headers.
> 
> I hope this helps,
> 
> Olaf
> 
> Sergio Queiroz wrote:
> > Hi!
> > 
> > I am just curious about the compilation time of ac++ 1.0, because it seems to
> > me that the older versions are something faster than the current one, tried to
> > compile some examples and it takes much time.
> > 
> > Am I correct about this feeling? I also saw the roadmap of next version and
> > there is an item "improved performance".
> > 
> > What I can do to reduce the compilation time?
> > 
> > Sérgio
> > 
> > _______________________________________________
> > aspectc-user mailing list
> > aspectc-user at aspectc.org
> > http://www.aspectc.org/mailman/listinfo/aspectc-user
------- End of Original Message -------

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Makefile
Type: application/octet-stream
Size: 296 bytes
Desc: not available
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: aspect.ah
Type: application/octet-stream
Size: 194 bytes
Desc: not available
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment-0001.obj>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: log092.txt
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment.txt>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: log1.0.txt
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment-0001.txt>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: main.cpp
Type: text/x-c++src
Size: 463 bytes
Desc: not available
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment.cpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test.cpp
Type: text/x-c++src
Size: 101 bytes
Desc: not available
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment-0001.cpp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test.h
Type: text/x-chdr
Size: 255 bytes
Desc: not available
URL: <http://www.aspectc.org/pipermail/aspectc-user/attachments/20051122/b069c2af/attachment.h>