Re: [Texmacs-dev] Cache profiling of TeXmacs 1.0.3.9

[David MENTRE]

Hello David,

David Allouche <address@hidden> writes:

>   1. Guile is crap, but we already knew that.

But we can give figures on the crapiness. :)

>   2. TeXmacs has many cache misses, but they seem to happen all over the
>      code. So, the issue are either general design problems, or are
>      faults in commonly used data structures.
>
>      Here, my usual suspects are needless indirections. They might make
>      not a big difference. But that's something which should be tried.

I think I can give figure on that. (to be continued)

>   3. The top faulters are counted pointer constructors and destructors
>      (not real objects ctors and dtors, just pesky counted pointers)
>      and the memory allocator.

ditto.

>   4. Another remarkable faulter is the string comparison.


  5. General memory consumption

     I agree with David that the more memory you consume, the slower is
     your program. You pay a penalty of 10 times to 100 times when going
     from L2 cache to main memory. The faster is your machine, the more
     you pay. So erasing of uneeded memory consumption and improving
     locality can improve performances.

     Of course, good algorithms are the first reason of performances.
     But apparently TeXmacs is good on that side (not counting the
     strictly increasing to do list of Joris ;-).


Yours,
d.

PS : David, diging into texmacs-dev archive, I found your post of
     texmacs+boehm gc but no source code. Is this work available
     somewhere? 
-- 
 David Mentré <address@hidden>