[mercury-users] Notes for a talk.

Fergus Henderson fjh at cs.mu.oz.au
Fri Oct 31 05:30:54 AEDT 1997

I wrote:

> Of the remaining 35% overhead, which is only 0.12 seconds
> (Mercury takes 0.47 seconds while C takes 0.35 seconds),
> I think that most of it (0.09 seconds or 25%) is start-up costs in
> the Mercury runtime.  I haven't analyzed the causes for this,

When I started analyzing the causes, I realized this was in fact incorrect.
Although the startup overhead was 0.09 seconds when using the conservative
collector, in the non-GC case startup overhead was only 0.01 seconds.

So I suppose all of the 35% overhead must be due to the causes noted earlier:
suboptimal use of registers and a few unnecessary loads and stores. 

Fergus Henderson <fjh at cs.mu.oz.au>   |  "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh>   |  of excellence is a lethal habit"
PGP: finger fjh at         |     -- the last words of T. S. Garp.

More information about the users mailing list