bad linking times [was Re: SUBMISSION: New Library for kdesupport: indexlib]

Michael Pyne pynm0001 at comcast.net
Sat Jun 11 02:39:42 BST 2005


On Friday 10 June 2005 20:43, Brad Hards wrote:
> On Sat, 11 Jun 2005 05:37 am, Michael Pyne wrote:
> > Even without --enable-final, kdebindings/smoke takes damn near forever
> > for me. Ironically it spends most (damn near an hour) of its time linking
> > (I suspect because I pass -Wl,-O1 to ld).
>
> A really new binutils might help - Ben Elliston told me that he fixed a
> terrible algorithm that made a lot of difference. Not sure which version,
> or even if it is applied.

I had 2.16 installed.  The only newer version I can find is a CVS pre-release 
that Gentoo has packaged (2.16.90.0.3).  I decided to try that to see if it 
was the difference.  I'm using juk to test instead of smoke to save me some 
time. ;-)

What I did was do "rm juk ; time unsermake" in the juk build directory.  
ccache doesn't apply to this step so you don't have to worry about that.  I 
performed this step once before recording times in order to eliminate cache 
issues.  The times listed are the total wallclock times (and I only did it 
twice for expediency).

2.16              2.16.90.0.3
--------          --------
11.92s            10.36s
11.79s            10.32s

So there is definitely an improvement, but it doesn't look like it would help 
all that much linking smoke (unless the change in question is one that e.g. 
changes an O(n^2) algorithm to O(n log n) and my n is too small to perceive 
the change).  But I don't want to delete libsmoke to find out. =D

Regards,
 - Michael Pyne
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20050610/7d704ea3/attachment.sig>


More information about the kde-core-devel mailing list