[Ecls-list] Upcoming optimization of slot accessors

[Juan Jose Garcia-Ripoll]

On Fri, Dec 16, 2011 at 10:39 PM, Juan Jose Garcia-Ripoll <
juanjose.garciaripoll at googlemail.com> wrote:

> Ok, the patches are now in git/CVS. Using the code below, compiled, these
> are the benchmark figures of FPRINT/PRETTY for different commits. The
> greatest improvement was the use of special dispatch functions.
>
> I have been doing some intensive checks of the functions, but please,
> should you find any problem, report it ASAP and I will try to fix it.
>
> Now, back to the cave for some more optimizations...
>

Here is one more (well, actually it is a fix for an optimizer bug!). 1.178
/ 3.445 ~ 34%

Fixed TYPEP optimizer
;;; Loading "/Users/jjgarcia/tmp/foo.fas"
real time : 1.214 secs
run time  : 1.178 secs
gc count  : 1 times
consed    : 14342976 bytes

 Starting point
> ;;; Loading "/Users/jjgarcia/tmp/foo.fas"
> real time : 3.534 secs
> run time  : 3.445 secs
> gc count  : 2 times
> consed    : 20413056 bytes
>
> SETF functions stored in compiled code
> ;;; Loading "/Users/jjgarcia/tmp/foo.fas"
> real time : 3.449 secs
> run time  : 3.250 secs
> gc count  : 2 times
> consed    : 20410992 bytes
>
> New dispatch for slot accessors
> ;;; Loading "/Users/jjgarcia/tmp/foo.fas"
> real time : 1.873 secs
> run time  : 1.836 secs
> gc count  : 2 times
> consed    : 20410896 bytes
>

-- 
Instituto de Física Fundamental, CSIC
c/ Serrano, 113b, Madrid 28006 (Spain)
http://juanjose.garciaripoll.googlepages.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/ecl-devel/attachments/20111217/efe1a866/attachment.html>