Comments on: Array

By: Mitch

Mitch — Sun, 25 Apr 2010 20:03:35 +0000

“switching data structure implementations around 30,000 elements” – shouldn’t this be 300,000 ?

By: david

david — Mon, 21 Aug 2006 11:35:33 +0000

dont understand your ‘number’ thing, but …. still your articles are very good!

By: david

david — Mon, 21 Aug 2006 11:34:00 +0000

Finally, i understood… thank you – very good article

By: Peteris Krumins

Peteris Krumins — Sun, 23 Jul 2006 09:39:18 +0000

Just to inform you, classes:
o hash_set
o hash_map
o hash_multiset
o hash_multimap

are not part of the C++ standard but are CGI extensions.

By: Raid Array » SuperArray - focused microarray, PCR, siRNA and Gene Function Assay

Mon, 22 May 2006 15:01:24 +0000

[...] ex of -1 indicates the last element of the array, -2 is the next to last ridiculous_fish Blog Archive ArrayCopyOnWriteArraySet is a set that implements copy-on-write and is ba [...]

By: Zac’s Weblog » Blog Archive » To CD or not to CD?

Zac’s Weblog » Blog Archive » To CD or not to CD? — Fri, 12 May 2006 18:30:22 +0000

[...] it back. You can see a movie of it working here. Because of Ridiculous Fish’s great article on reverse engineering Apple’s NSArray implementation, I also think that [...]

By: Patrick Stein

Patrick Stein — Sun, 26 Mar 2006 20:36:06 +0000

Hmmmm timing for the plain array seems a bit weird. A plain array that backward walks grows – isn’t it just a

avalue = anarray[offset] ?

so in real just an addition of addresses:

*(anarray+offset) ???

with offset just decreasing ? If you mean that walking through the array takes longer I do undrestand that, but how could that get faster with other data structures ?

Enligthen me – Patrick – aka Jolly

By: Daniel Berlin

Daniel Berlin — Mon, 16 Jan 2006 23:13:47 +0000

So, what is the memory behavior of CFArray like (ie memory usage)?

The only thing it seems truly better at is deletes, and the only way i can imagine it could remain flat like that for random deletions is to not do the actual deletes, but just mark them to be deleted. This would explain the walk timings as well.

By: Rosyna

Rosyna — Mon, 16 Jan 2006 21:33:04 +0000

Have you thought about doing the same benchmark with NSArray that you did with CFArray? I mean, just to see the overhead that Objective-C adds.

By: nicolas

nicolas — Fri, 30 Dec 2005 10:27:22 +0000

thanks zac for your reply !
I’d like to add one observation.
It may seem obvious but it can’t hurt to repeat it :
The biggest optimization I’ve made on an app that used sorted arrays was : limiting the amount of times I sorted the damn arrays !
I spent some time trying to optimize the sorts or the sorted insertions (binary insert etc). This gained very little cycles…. But calling the sort functions every 100ms instead of every 25ms made a HUGE difference ^_^