glob2-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [glob2-devel] new gradient system


From: MOA333
Subject: Re: [glob2-devel] new gradient system
Date: Wed, 18 Aug 2004 12:38:35 +0000
User-agent: Mozilla Thunderbird 0.7.3 (X11/20040812)

Nuage wrote:

MOA333 wrote:
| did you saw the tests?
|   the full recursive checkSums is taking 60% and more of CPU time
| on 256*256
|
|   Do you think we can do somethink about this?

We could very easyly remove it, but this would make the desynchronisation
bugs harder to find. All this CPU is used into Map, Sint32 checkSum(bool heavy);
when heavy=true. I think that we'll add an option to enable/disable it.


| (still the gfx on 128*128 but there i have lmess than 50% of cpu load so
| i guess it dosent matter)

I'd like to know the cpu you made thoses tests on. Especially what are
the size of the cache (level 1 and 2) of you cpu.


| why it is so big diference betwen 128*128 and 256*256?

Because some computing is propostional to the surface of the map.
The gradient system is, and Map::checkSum(heavy) is when heavy=true.
And a 256x256 map is *four* time bigger than a 128x128, so if thoses
computation time are the mosr important, it is 4 times slower!

i have the Athlon XP 2500+ with 512 cache i guess, not sure, and 128 or 256 level 1 but nor very surre about, anyway there is only one 2500+ i dont remember his cache.....
_______________________________________________
glob2-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/glob2-devel






reply via email to

[Prev in Thread] Current Thread [Next in Thread]