[Snippet] BitInt

[Nestharus]

This resource has moved here.

--

gone

 

[Zwiebelchen]

So basicly all you did is outsourcing BitManip as a requirement?

Anything new compared to the old BigInt? Where's all the maths stuff?


EDIT: Wow, my bad... I totally read BigInt, not BitInt ...

However, is this faster than using BigInt to store data? As its completely binary, I suppose not.

 

[Nestharus]

This is completely different from BigInt ; )

bits are written, so you don't multiply and add like you did before, you just write bits

because bits are written, the size is going to bloat a bit. The multiply in BigInt lets you store with maximal accuracy, down to fractions of bits. When you just use bits directly, you lose that accuracy, thus bloat. However, BitInt is MUCH faster than BigInt and as such can handle MUCH more data w/o freezing or thread crashing. The operation magnitudes are also independent from the number of bits in a BitInt, unlike BigInt, whose operations grow as the BigInt grows.

You want to use BigInt when you are dealing with standard save/load codes and BitInt when you are doing save/load with networking.

 

[Zwiebelchen]

Thanks for the quick explanation.

You want to use BigInt when you are dealing with standard save/load codes and BitInt when you are doing save/load with networking.

... hmm, so basicly this only makes sense when used in combination with hostbots or website-hosted games?

 

[Nestharus]

no...

http://www.hiveworkshop.com/forums/spells-569/file-i-o-2-0-0-1-a-217403/
http://www.hiveworkshop.com/forums/spells-569/network-v1-0-0-0-a-226065/

 

[Dr Super Good]

Any more revision tiers and it will be incrementing every time you think about making changes.

Honestly, why 4 tiers of revision?

 

[Nestharus]

Added 4 new methods

*           //dangerous, read note
*           method pushFront takes BitInt bits returns nothing
*               -   Pushes bits on to front of this
*               -   Can push on to partials, but can't push parials
*               -   A BitInt is partial if bitCount%bitGroup != 0
*
*           //dangerous, read note
*           method pushBack takes BitInt bits returns nothing
*               -   Pushes bits on to back of this
*               -   Can push partials, but can't push on to partials
*               -   A BitInt is partial if bitCount%bitGroup != 0
*
*           method popFront takes integer nodeCount returns BitInt
*               -   Pops off front nodes and returns BitInt containing them
*
*           method popBack takes integer nodeCount returns BitInt
*               -   Pops off back nodes and returns BitInt containing them

 

[Nestharus]

Added new function and provided crash safety on bit group change

it can process 550 digits per trigger evaluation, which is much faster than BigInt could ever do

No need for convertString safety as that's ridiculously fast

 

[Nestharus]

Updated convertString. It wasn't bugged, but it produced strange behavior when passing in strings with leading 0s.

 

[Nestharus]

Made deallocate O(1)

 

[Almia]

Is this the one you used for compressing integers?

 

[Nestharus]

huh? I use BigInt or BitInt depending on the situation

 

[ScorpioT1000]

math? no, never heard...

// cJass
#define IntGetByte1(src) = (src / (256*256))
#define IntGetByte2(src) = (ModuloInteger(src, (256*256)) / 256)
#define IntGetByte3(src) = (ModuloInteger(src,256))
#define IntSetByte1(src,val) = { src = (ModuloInteger(src,(256*256)) + val*256*256 ) }
#define IntSetByte2(src,val) = { src = (ModuloInteger(src,256) + (src / (256*256))*256*256 + val*256) }
#define IntSetByte3(src,val) = { src = ((src / 256)*256 +val)}

Hello from Russia

 

[Almia]

Cool!
Though there has to be a reason why that shouldn't be valid.

This system was used in CSL for bits.

 

[Magtheridon96]

math? no, never heard...

// cJass
#define IntGetByte1(src) = (src / (256*256))
#define IntGetByte2(src) = (ModuloInteger(src, (256*256)) / 256)
#define IntGetByte3(src) = (ModuloInteger(src,256))
#define IntSetByte1(src,val) = { src = (ModuloInteger(src,(256*256)) + val*256*256 ) }
#define IntSetByte2(src,val) = { src = (ModuloInteger(src,256) + (src / (256*256))*256*256 + val*256) }
#define IntSetByte3(src,val) = { src = ((src / 256)*256 +val)}

Hello from Russia

That's not relevant to this resource and the use of ModuloInteger would get us closer to the operation-limit. (Meaning MD5 and AES wouldn't work well :v)
Also, that code has the wrong behavior for negative numbers.

 

[Nestharus]

@ScorpioT1000

Is this what you were looking for?

http://www.hiveworkshop.com/forums/jass-resources-412/snippet-bit-manip-226117/

 

[ScorpioT1000]

Just wanna caution users about low perfomance due to unnecessary loops, nested calls etc. when you can use fast x86 processor math operators and cJass defines instead.

 

[ScorpioT1000]

That's not relevant to this resource and the use of ModuloInteger would get us closer to the operation-limit. (Meaning MD5 and AES wouldn't work well :v)
Also, that code has the wrong behavior for negative numbers.

I didn't create SetByte4 but you can if you want, this is not too difficult.

To solve oplimit you can use 1 periodic timer with custom stack (ExecuteFunc leads to desync, tested), this is not a problem at all.

 

[Magtheridon96]

It would be a hassle.

I'd rather just have the ability to call MD5(<some array of bytes>) and get the hash immediately :/

 

[ScorpioT1000]

If immediately you mean n seconds, ok

 

[Nestharus]

huh?

when you pack the BitInt in bytes, it's an O(n) loop, no math involved...

before I do MD5/AES calculations, I pack in bytes

 

[ScorpioT1000]

Magtheridon96, ok, I meant the oplimit resolution should be incapsulated into MD5 system and you can call MD5 without worrying about threads.

Nestharus, the question is in perfomance of your overheaded OO according binary operations. Nobody in pro dev does such kinda things. But in pro langs there are good optimizators and directives (like marco and inline) for such kinda calls, what I cannot say about vjass =)