Jump to content
  • 0

Chance Skills Again


Question

12 answers to this question

Recommended Posts

  • 0
Posted

Because double got similar performance on nowadays CPU. And chating about memory consumption, on maybe 200 double or short (we speak about chance skill), is pointless.

 

Why not short instead of int ? There are plenty of locations where integers could be replaced. You won't replace it because it doesn't worth the change.

  • 0
Posted

Because double got similar performance on nowadays CPU. And chating about memory consumption, on maybe 200 double or short (we speak about chance skill), is pointless.

 

Why not short instead of int ? There are plenty of locations where integers could be replaced. You won't replace it because it doesn't worth the change.

 

so lets go long all the way and forget about int, short or byte. Just because it works

  • 0
Posted (edited)

Type "double vs float" on Google, you will see most topics on wherever websites recommend to use directly double.

 

And you didn't answer my second question. There are places on the code using more expensive datatypes, I give you an emblematic exemple and you will maybe stop with that float stuff : L2World is cutted into L2WorldRegion, which are 88 * 128 = 11264 L2WorldRegion.

 

Each L2WorldRegion weights :

- 2 ConcurrentHashMap

- 2 ArrayList

- 2 int

- 1 ScheduledFuture

- 1 boolean

 

Now if I tell you than off those 11264 L2WorldRegion, 2368 L2WorldRegion got no reason to exist (which is atm 21%), because those areas got no geodata and therefore shouldn't be accessible by player/monster, but are still currently filled (since L2J exists). Still arguing about float vs double ?

 

You will make more profit decreasing L2WorldRegions putting given invalid regions off limits, than editing all possible datatypes to be "memory efficient".

 

Finally most L2 packets use int, and not long. So nope you will make a bad move using long :P.

Edited by Tryskell
  • 0
Posted (edited)

 

gameserver.model.ChanceCondition

private final int _chance;

i just need to change it to 

private final double _chance;

?

Edited by ElTyto
  • 0
Posted

Type "double vs float" on Google, you will see most topics on wherever websites recommend to use directly double.

 

And you didn't answer my second question. There are places on the code using more expensive datatypes, I give you an emblematic exemple and you will maybe stop with that float stuff : L2World is cutted into L2WorldRegion, which are 88 * 128 = 11264 L2WorldRegion.

 

Each L2WorldRegion weights :

- 2 ConcurrentHashMap

- 2 ArrayList

- 2 int

- 1 ScheduledFuture

- 1 boolean

 

Now if I tell you than off those 11264 L2WorldRegion, 2368 L2WorldRegion got no reason to exist (which is atm 21%), because those areas got no geodata and therefore shouldn't be accessible by player/monster, but are still currently filled (since L2J exists). Still arguing about float vs double ?

 

You will make more profit decreasing L2WorldRegions putting given invalid regions off limits, than editing all possible datatypes to be "memory efficient".

 

Finally most L2 packets use int, and not long. So nope you will make a bad move using long :P.

 

 

Who fucks the performance of float vs double or short vs long, they are either 32bit on a 32 lenght bus or 64 bit (long) on a 64 bit bus and yes you neither benefit in memory consumption.

 

Then why use final if its only purpose is to help the developer know that this variable won't change (no perfomance gain runtime), why use generics if they are just precompiled and dont exist after compilation that also slows down due to lots of casts under the hood. Why use void methods if at the end theres always a return value (garbage or not there actually is.)

 

Its not a matter or perfomance, its a matter of good, reusable and understandable code. On this case does float vs double make any difference at all? No but you have to start from somewhere

  • 0
Posted (edited)

Because double got similar performance on nowadays CPU. And chating about memory consumption, on maybe 200 double or short (we speak about chance skill), is pointless.

 

Why not short instead of int ? There are plenty of locations where integers could be replaced. You won't replace it because it doesn't worth the change.

that information was VERY, VERY IMPORTANT.

I use float in my changes, now I'll update all to double! :) (Edited to be like Tryskell suggested below)

 

Thanks 

Edited by colt
  • 0
Posted (edited)

if you already used float, keep float. But using float instead of double - until you use millions entries - is simply pointless. There are non optimized places all over sources (I gave a single exemple which probably can save easily 5mo RAM if reworked) which got a WAY BIGGER IMPACT if reworked rather than replacing datatypes for the best datatype possible.

 

Ofc using the shortest datatype possible will be the most avantageous, but if you do it for a story of RAM, MOREOVER ON SOMETHING LIKE 100 ENTRIES, then it's pointless.

 

Remove a single int from L2Spawn, and you can see a change, as there are 40k NPC entries. But stating about float<>double on chance skills... Blurp.

 

And it's "double", not "Double". "Double" is an object, "double" is a primitive. "Double" costs heavier than "double". Same case for Boolean/boolean, Integer/int, etc.

Edited by Tryskell
  • 0
Posted

if you already used float, keep float. But using float instead of double - until you use millions entries - is simply pointless. There are non optimized places all over sources (I gave a single exemple which probably can save easily 5mo RAM if reworked) which got a WAY BIGGER IMPACT if reworked rather than replacing datatypes for the best datatype possible.

 

Ofc using the shortest datatype possible will be the most avantageous, but if you do it for a story of RAM, MOREOVER ON SOMETHING LIKE 100 ENTRIES, then it's pointless.

 

Remove a single int from L2Spawn, and you can see a change, as there are 40k NPC entries. But stating about float<>double on chance skills... Blurp.

 

And it's "double", not "Double". "Double" is an object, "double" is a primitive. "Double" costs heavier than "double". Same case for Boolean/boolean, Integer/int, etc.

Thanks again, awesome (I print to pdf for documentation purpose!) :)

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



×
×
  • Create New...