mirror of https://github.com/CGAL/cgal
109 lines
5.1 KiB
Plaintext
109 lines
5.1 KiB
Plaintext
[See also the file TODO_static_filters]
|
||
|
||
- Filtered_kernel<> should be a proper kernel. It should be possible to use
|
||
the Kernel_base stuff (see Kernel_traits paper)...
|
||
- mv test/include/*.C *.def ; Or use template functions ?
|
||
|
||
Concerning the main code:
|
||
-------------------------
|
||
- Policy for overlapping comparisons : another good (faster) solution
|
||
would be to set a static boolean variable to indicate a buggy comparison
|
||
happenned (a la IEEE inexact flags). Drawback is that it's not thread safe,
|
||
but could be made safe with pthread_key_create()...
|
||
Clearly, all this stuff is policy.
|
||
struct overlap_throw {
|
||
void operator() const { throw(ce_qui_faut); }
|
||
};
|
||
Or :
|
||
class overlap_static {
|
||
static bool pipo = false;
|
||
public:
|
||
void operator() const { pipo = true; }
|
||
void reset() const { pipo = false; }
|
||
};
|
||
- Pentium 4 : gcc -march=pentium4 -mfpmath=sse supports something like:
|
||
#include <xmmintrin.h>
|
||
#include <iostream>
|
||
|
||
int main()
|
||
{
|
||
int i = _mm_getcsr();
|
||
std::cout << i << std::endl;
|
||
|
||
x = _MM_GET_ROUNDING_MODE()
|
||
_MM_SET_ROUNDING_MODE(x)
|
||
return 0;
|
||
}
|
||
Faut juste bien faire gaffe a __SSE2__ pour savoir quelle unite est utilisee
|
||
pour les calculs (x87 ou SSE2). Ca pourrait meme etre les deux (bonjour le
|
||
bordel...).
|
||
Avantage : y'a pas de risque d'overflow/underflow avec la SSE2, donc on peut
|
||
disabler le workaround.
|
||
Y'a peut-etre meme des instructions interessantes pour speeder l'IA avec la
|
||
SSE2.
|
||
- Use expression templates for Interval_nt<> and Lazy_exact_nt<>.
|
||
- Specializations for is_zero() and co, and make use of them in the kernel.
|
||
[Filtered_exact<> doesn't define them either, it's a bug...]
|
||
- SunPro 5.x supports Interval arithmetic...
|
||
http://docs.sun.com/htmlcoll/coll.693.1/iso-8859-1/CPPARITHPROG/iapg_bookTOC.html
|
||
http://www.sun.com/forte/cplusplus/interval/index.html;$sessionid$GSXJDFYAABWVTAMTA1FU45Q
|
||
- Try to get rid of the libc5 compatible version (make benchmarks first).
|
||
- Have determinant_by_formula() overloading for Lazy_exact_nt<> ?
|
||
And what about for Interval_nt<> too ?
|
||
- Turn CGAL_IA_CHECK_RESTRICT into an [expensive] assertion ?
|
||
- Get rid of convert_to() ? This is a mess for nothing (benchmark first).
|
||
Target convert_from_to (const Target, const Source) is fine ?
|
||
It could maybe be made better this way:
|
||
void convert_from_to(Target &, const Source &) ?
|
||
Suggest that to Stefan for misc.h ?
|
||
- Handle in_smallest_orthogonalcircle_ftC2.h correctly (needs an include)
|
||
- Bench -fbranch-probabilities ? Use __builtin_expect() for GCC 3 ?
|
||
- Mark the cache as "mutable" (see Stroustrup, page 232) ?
|
||
- Filter_Cache: Faire des benchs, et une test-suite qui soit raisonnable.
|
||
Hum, rajouter un bool<6F>en pour calculer le cache seulement sur demande ?
|
||
(<28>a <20>vite de le faire inutilement pour les variables interm<72>diaires,
|
||
mais <20>a prend un chouia plus de place... mais en comparaison du reste...)
|
||
- Replace CGAL_IA_MAX_DOUBLE by standard DBL_MAX in <cfloat>, if portable
|
||
(add a test). Not possible for CGAL_IA_MIN_DOUBLE, since DBL_MIN is the
|
||
_normalized_ minimum.
|
||
- See the C++ Standard numeric_limits<>, section 18.2.
|
||
? See ISO C99 and http://http.cs.berkeley.edu/~fateman/fp98/korenF/node3.html.
|
||
- "C9x FP unordered compares":
|
||
+ /* ISO C99 IEEE Unordered compares. */
|
||
+ builtin_function ("__builtin_isgreater", default_function_type,
|
||
+ BUILT_IN_ISGREATER, BUILT_IN_NORMAL, NULL_PTR);
|
||
+ builtin_function ("__builtin_isgreaterequal", default_function_type,
|
||
+ BUILT_IN_ISGREATEREQUAL, BUILT_IN_NORMAL, NULL_PTR);
|
||
+ builtin_function ("__builtin_isless", default_function_type,
|
||
+ BUILT_IN_ISLESS, BUILT_IN_NORMAL, NULL_PTR);
|
||
+ builtin_function ("__builtin_islessequal", default_function_type,
|
||
+ BUILT_IN_ISLESSEQUAL, BUILT_IN_NORMAL, NULL_PTR);
|
||
+ builtin_function ("__builtin_islessgreater", default_function_type,
|
||
+ BUILT_IN_ISLESSGREATER, BUILT_IN_NORMAL, NULL_PTR);
|
||
+ builtin_function ("__builtin_isunordered", default_function_type,
|
||
+ BUILT_IN_ISUNORDERED, BUILT_IN_NORMAL, NULL_PTR);
|
||
[voir le draft C99 ce que c'est]
|
||
? SUN: Add "-Wa,-xarch=v8plus" to fix the following "error":
|
||
/usr/ccs/bin/as: "/var/tmp/ccBHnU0T.s", line 861: warning: 3 instructions
|
||
required between ldfsr and next FBfcc; nops inserted
|
||
|
||
Concerning the script:
|
||
----------------------
|
||
- Produce separate files for declarations and definitions (=> libCGAL).
|
||
|
||
Concerning the doc:
|
||
-------------------
|
||
- In the 2.0 HTML doc, my enums are indexed twice.
|
||
probably a cc_manual compliance bug from me.
|
||
Idem, my fct to_double(Ia) is not the same as the others...
|
||
- add a pointer to my MISC'99 paper.
|
||
- DOCUMENT the new boolean template parameter, the script, the static filters.
|
||
|
||
Concerning the test-suite:
|
||
--------------------------
|
||
- Check it with GCOV again before the next public release.
|
||
- Make a more extensive test-suite for the filtered predicates.
|
||
The script could output information to test them generically somehow.
|
||
- Test NaN propagation. Comparisons with these should throw the exception...
|
||
Check that they are correctly propagated (by min(), max(), even operator*...)
|