[bug #68121] convert isequal to built-in function for performance

"Dmitri A. Sergatskov" <[email protected]> Mon, 4 May 2026 19:47:05 -0400 (EDT)
Newsgroups	gmane.comp.gnu.octave.bugs
Message-ID	<[email protected]>
Please use the bug tracker to post updates to a bug report.  The mailing list is intended as a read-only notification stream.  Info posted to this mailing list address won't appear in the tracker database where it is most useful.

Follow-up Comment #29, bug #68121 (group octave):

Here is  my proposal.
(file isequal_vmethod_20260504_200.diff)
it implements `isequal()` as a virtual method, so lots of files get touched. 
With attached benchamrk 
(file bench_isequal.m) I see (on Ultra 9 285 CPU) before:

octave:1> bench_isequal
empty loop (1e6)             : 0.0225 s
isequal(1,1) (1e6)           : 15.9582 s
double 5000x5000  equal x50   : 0.8293 s
double 5000x5000  differ x50  : 0.7901 s
single 5000x5000  equal x50   : 0.4861 s
int32  5000x5000  equal x50   : 0.4847 s
logical 5000x5000 equal x50   : 0.2531 s
complex 5000x5000 equal x50   : 1.3389 s
char   5000x5000  equal x50   : 0.0659 s
sparse 5000x5000  equal x200  : 0.7049 s
isequaln 5000x5000 w/ NaN x50 : 2.1113 s
cell 1x10000 (scalars) x100 : 17.3372 s
struct 3x500x500 x200     : 0.0705 s

After:

octave:1> bench_isequal;
empty loop (1e6)             : 0.0223 s
isequal(1,1) (1e6)           : 0.8347 s
double 5000x5000  equal x50   : 0.5819 s
double 5000x5000  differ x50  : 0.5240 s
single 5000x5000  equal x50   : 0.2621 s
int32  5000x5000  equal x50   : 0.2580 s
logical 5000x5000 equal x50   : 0.0664 s
complex 5000x5000 equal x50   : 1.0598 s
char   5000x5000  equal x50   : 0.0626 s
sparse 5000x5000  equal x200  : 0.2884 s
isequaln 5000x5000 w/ NaN x50 : 0.6700 s
cell 1x10000 (scalars) x100 : 0.0240 s
struct 3x500x500 x200     : 0.0473 s


Also some benchmarks previously on this report:

octave:2> M = randi([1 10],9)-1;
octave:3> tic,for idx = 1:100, M==M;endfor,toc
Elapsed time is 5.48363e-05 seconds.
octave:4> tic,for idx = 1:10000, all(M(:));endfor,toc
Elapsed time is 0.0139458 seconds.
octave:5> tic,for idx = 1:10000, isequal(M,M);endfor,toc
Elapsed time is 0.0095439 seconds.



octave:6> s1 = ['a', 'b'; 'c', 'd']
s1 =

ab
cd

octave:7> s2 = ['a', 'b'; 'e', 'f']
s2 =

ab
ef

octave:8> isequal(s1,s2)
ans = 0



octave:9> a = randi(250, 1e4, 1e4, "uint8");
octave:10> b = a + uint8(1);
octave:11> c = b - uint8(1);
octave:12> tic; isequal(a, b), toc
ans = 0
Elapsed time is 3.69549e-05 seconds.
octave:13> tic; isequal(a, c), toc
ans = 1
Elapsed time is 0.00641489 seconds.


Dmitri.
-- 


(file #58549, file #58550)

    _______________________________________________________

Additional Item Attachment:

Name: isequal_vmethod_20260504_200.diff Size: 62KiB

<https://file.savannah.gnu.org/file/isequal_vmethod_20260504_200.diff?file_id=58549>

Name: bench_isequal.m                Size: 3.7KiB
    <https://file.savannah.gnu.org/file/bench_isequal.m?file_id=58550>


    AGPL NOTICE

These attachments are served by Savane. You can download the corresponding
source code of Savane at
https://savannah.gnu.org/source/savane-c36938be85ff6c1b727bc7dd7fd30e48f9142870.tar.gz


    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?68121>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc (application/pgp-signature, 228 B)