Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Mego]

The most complicated way of getting random numbers yet, did render me some cool element and mineral names:

Code: [Select]

int NMaterialManager::getRandomInt(const int& seed, const int& max){
  boost::mt19937 gener(seed);  // Mersenne twister
  boost::uniform_smallint<> normal(0,max-1); // uniform distribution optimised for small integer ranges
  boost::variate_generator<boost::mt19937&,boost::uniform_smallint<> > rng(gener, normal); // Variatic generator combining the two
  return rng(); // Finally, a number.
}


Spoiler (click to show/hide)

galsunstarine
nitraastroate
astroate
sylon
rosesum
barbaride
neonitstarium
zeoalcit
alvite
berum
hyperpharmatite
pharmatite
currhodthulium
carhedobstone
helxenoite
oliarmokix
erbine
hessbiorock
biocascobix
conerbkryptium
elstone
teraxenobaltium
graphmesohydroaplith
mesohydroapsylophane
sylthulrubine
mixapbersite

Damn. Imagine trying to cram all of that into one line.

[Siquo]

Damn. Imagine trying to cram all of that into one line.

Assuming that was sarcasm:
No premature optimisation! Writing that into one line would just mean re-writing it again when I make them resettable members instead of re-allocating it for every call. I'm just getting into boost and learning what's in there. It's a lot...

[Skyrunner]

Deciding on a point between memory and performance is hard. Dx Right now, I'm dumping efficient memory usage, in having every variable stored instead of being looked up. I don't know if it's a good idea, but I'm really feeling that trade-off. Ten megabytes of ram!

[Max White]

What are you trying to do, Skyrunner? It might be that flyweight pattern helps you out here, I don't know, depends on the problem.

[Skyrunner]

Oh, hmm. I could see how I could make that work, if each variable is replaced with a pointer to a data structure that has the actual information. It would reduce memory usage by a lot.

I feel too lazy to do that... maybe after I make a working demo. xD

[Max White]

I feel too lazy to do that

The second most common cause of people not using OOP when it will help them. The first most is not knowing what to do. 

I'll just sit in my corner, rambling about patterns...

[Skyrunner]

...But I am using OOP. The flyswatter pattern is just a pattern, not OOP embodied. xD

Besides, I'm prioritizing! Ten megabytes on a four gigabyte computer won't matter right now! xD

[Max White]

Grr... Kids these days! *Grumble* give you information expert! * Mumble mumble* so much force you will feel it through your facade!!!!

 

[MagmaMcFry]

Extremely. It's difficult even getting through the opening menu, and the game is entirely unplayable.

The fact that the opening MENU almost locks up is a preeeetty good sign there's something wonky going on in your code.

That's extremely weird. On my laptop, it uses only a tiny fraction of CPU/GPU time/memory, yet it still runs at 60 FPS without any problems, even when I scale the game speed to 100 ticks/frame (instead of the 1 tick/frame that I uploaded the game with). I have loader classes that make sure that no external file is loaded twice. Yet you still experience lag? I'll try changing the rendering approach, let's see if that works.

[olemars]

Extremely. It's difficult even getting through the opening menu, and the game is entirely unplayable.

The fact that the opening MENU almost locks up is a preeeetty good sign there's something wonky going on in your code.

That's extremely weird. On my laptop, it uses only a tiny fraction of CPU/GPU time/memory, yet it still runs at 60 FPS without any problems, even when I scale the game speed to 100 ticks/frame (instead of the 1 tick/frame that I uploaded the game with). I have loader classes that make sure that no external file is loaded twice. Yet you still experience lag? I'll try changing the rendering approach, let's see if that works.

Is this a 3D app? Perhaps with shaders?

[Max White]

I don't suppose you are running any weird dev version of Java that might still be prone to these sort of problems, Glyph?
It seems strange that my laptop can handle something that a real computer can't, so it seems very unlikely to be eating resources.

[MagmaMcFry]

Is this a 3D app? Perhaps with shaders?

It does use OpenGL polygons instead of blitting, but no shaders.

[MorleyDev]

The most complicated way of getting random numbers yet, did render me some cool element and mineral names:

Code: [Select]

int NMaterialManager::getRandomInt(const int& seed, const int& max){
  boost::mt19937 gener(seed);  // Mersenne twister
  boost::uniform_smallint<> normal(0,max-1); // uniform distribution optimised for small integer ranges
  boost::variate_generator<boost::mt19937&,boost::uniform_smallint<> > rng(gener, normal); // Variatic generator combining the two
  return rng(); // Finally, a number.
}


You're reseeding the generator every time it is called. Both pointless and possibly expensive with Mersenne's  twister, just a heads up :-)

[Siquo]

You're reseeding the generator every time it is called. Both pointless and possibly expensive with Mersenne's  twister, just a heads up :-)

No and yes. I know it's expensive, but not pointless: I need to have the same results every seed (pseudorng), and I vary the seed myself. I know the Mersenne isn't a really good choice for this, if you know a better one (cheaper to re-seed), I'd like to know

So, in my simulation, I have a seed number, that renders a Material Name, which is then re-converted into a number, which is then used to seed the properties of that material. So every time and everywhere you encounter a material called "hyperpharmatite", it will have the exact same properties.

[Max White]

I like what Siquo is doing.
It isn't a random number generator, it is a hard to predict sequence generator.