Home » Extra libraries, Code snippets, applications etc. » C++ language problems and code snippets » Inverse palette conversion algorithm...
Re: Inverse palette conversion algorithm... [message #2389 is a reply to message #2244] |
Mon, 10 April 2006 17:58   |
mr_ped
Messages: 826 Registered: November 2005 Location: Czech Republic - Praha
|
Experienced Contributor |
|
|
I did some benchmark on "old" Athlon 1GHz, and your old code is almost 3 times faster than my "supposed to be faster" code!
(original benchmark with my code being 2 times faster was from Celeron 2GHz)
That's ridiculous, I think that can't be the data cache trashing anymore, but I didn't have time to work on it more and figure out what's happening.
But having such simple code so significantly faster/slower on different CPU did really surprise, I was expecting like 50% speed differences between AMD vs Intel and different cache sizes.
Anyway, I did have time to *think* about it, and I can assure you the code I posted is just a mere "proof of concept". I didn't had time to try all those optimisations yet if they are correct - but if they are, expect some surprises and some more coding tricks. (and IMHO really nice ones )
Especially the above mentioned bottleneck of large Vector[] table will be removed to the extend I didn't even dream it was possible. 
I think I will be back with improved version in 2-3 days, when I will have some time to sit behind my home PC and try all those ideas.
So far I have a little programmer's quiz here:
How will I make the "feed_me" array about 100 times smaller without any really "dirty" trick or major algorithm change? 
(the code will change considerably, but the basic idea of the whole code will remain same)
|
|
|
 |
|
Inverse palette conversion algorithm...
By: mirek on Wed, 05 April 2006 10:11
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Fri, 07 April 2006 10:05
|
 |
|
Re: Inverse palette conversion algorithm...
By: mirek on Fri, 07 April 2006 11:16
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Fri, 07 April 2006 13:23
|
 |
|
Re: Inverse palette conversion algorithm...
By: mirek on Fri, 07 April 2006 13:34
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Fri, 07 April 2006 13:16
|
 |
|
Re: Inverse palette conversion algorithm...
By: mirek on Fri, 07 April 2006 13:33
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Fri, 07 April 2006 17:30
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Fri, 07 April 2006 17:46
|
 |
|
Re: Inverse palette conversion algorithm...
By: mirek on Sat, 08 April 2006 14:00
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Mon, 10 April 2006 17:58
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Sun, 21 May 2006 15:40
|
 |
|
Re: Inverse palette conversion algorithm...
By: mr_ped on Sun, 21 May 2006 15:46
|
 |
|
Re: Inverse palette conversion algorithm...
By: mirek on Sun, 21 May 2006 16:01
|
Goto Forum:
Current Time: Tue Jul 08 01:11:19 CEST 2025
Total time taken to generate the page: 0.03401 seconds
|