Legacy GM Binaries / Bytes ... performance and structural optimization.

[Khanon]

Spoiler: codes_factory

statut : *under construction*............................................version_new : 0.00 (20/03/18)
progression : 000.01%....................................................version_old : 0.00 (15/03/18)
usability : 000.00%..........................................................bonus code added : 0
notice : no 3d, no internet will be concerned
//======================================
protocole :
create a new project | name it as you like
put room0 background_color to black | set the room0 dimension to ?? px height and ?? px width
create 0 sprite
create 0 script
create 1 font | set font0 to text_size 22
create 2 new objects
put object0 anywhere in the room
//======================================

Spoiler: object0

//_create_event
a=0
for (i=0 ; i<2 ; i++) { create_instance(x,y+40*i,object1) }

Spoiler: object1

//_create_event
a=0

//_draw_event
draw_text(x,y,a)

Spoiler: topic

Hi, i open a topic on binaries.

i try to figure out how the system variables (and not the variables) handle binary transferts
... and by extension, how works the binaries concatenation in tcp packets (aproximate approach).
i don't want to know how to code binary because i already know.
what i leak is the ideal variable structure for maximize informations and performance (binary) transferts.
​

So this subject is more conceptual than it appears, i will try to condensate all my knowledge and shared point of views with you in this first post. Hope it will deserve a good reflection about it.

0) Basics
.........1.0 byte = 8 bits
.........0.5 byte = 4 bits
.........~.1 byte = 1 bit
Variables :
....are used to store information in the devices memory for later (or instant) use.
....can store either a real number or a string.
....have one name with a maximum length of 64 symbols.

1) Recaptitulation
>>> In GML
Variables are all stored as (64-bits ?) double-precision floating point numbers, (...).====> 8 bytes (floating 4 bytes ?)
Real numbers are all stored as 32-bits floating point values (even integer values), (...). => 4 bytes (floating 8 bytes ?)

>>> in Mysql
The maximum combined size of all BIT columns used must not exceed 512 bytes (of 8 bits per byte).
4 bytes | INTEGER, INT....................................1 byte- | TINYINT
4 bytes | FLOAT................................................2 bytes | SMALLINT
8 bytes | FLOAT................................................3 bytes | MEDIUMINT
8 bytes | DOUBLE............................................8 bytes | BIGINT
8 bytes | REAL

2) Questions and Interpretation
=> When you work with values (like toto+=1) ...
....about the increment " 1 " ... is it a 1 byte process ? an half-byte process ? or a mono-bit process ?
....about the variable " toto "... is this variable is also byte processing with the incremented value ?
....about bytes consumption ... are they byte floored to 1 above ? or halves-bytes are also accounted ?

Spoiler: old questions

* empty*

3) For tcp ... later on

Thanks for any help about more information on this subject.
//=============================================================
Last update (20/03/18) <= (17/03/18) <= (16/03/18) <= ...

Spoiler: feeders

yoyogames_help_manual
Tthecreator
dev_Mysql_docs

If you reached the maxima integer value > go to decimal (int 1 > dec 1.0)

 

[Tthecreator]

Unfortunately things are a bit more complicated than that.
In GM, all real variables are doubles. This is a special datatype that has 2 seperate parts: the number and the precision of that number.
It's kinda like scientific notation like: 5.6 * 10^7.
This allows you to have very small precise numbers with many decimal places, or a huge number without decimal places.
So it's processed quite weird and I don't know exactly how but I'm guessing that the entire variable would have to be loaded into the cpu.

I think this is what you were asking. I don't have any clue what you want to do with TCP though?

 

[Simon Gust]

Here's another fact: Any bitwise operation does convert the value to a int64, also works in reverse.

 

[Yaazarai]

This is very oddly worded, but I think I kind of see what you're asking. So basically how does GameMaker handle variables when storing strings, numbers, etc.? And how does that affect writing and reading packets in TCP protocol?

GameMaker uses 3 fundamental types as far as I know: string, 64-bit float and 64-bit signed integer. GameMaker uses something similar to a generic type, you give the variable an input and GameMaker determines if the input is one of the three afformentioned types. GameMaker can do bit-wise operations on the 64-bit integer type and they work as expected (except on GMS 1.4 and prior where GM would convert to 32-bit integers on bit-wise operations).

However if you perform a bit-wise operation on a float GameMaker will truncate the decimal places and with do operations on the integer part of the number. For example if I do a left bit-shift on a float this is what'll happen:

var float = 1024.34098;
float = float << 2;
// RESULT: 4096

var string = "HELLO";
string = string & 45;
// RESULT: 0

As you can see the floating decimal 0.34098 was truncated in favor of the integer part of the number as expected. And on the string the value was ismple set to 0.

As far as how does this affect TCP protocol: Type converstion. GameMaker will cconvert from the string, 64-bit float or 64-bit int over to whatever type you want to put in a buffer and send off over your TCP protocol and GameMaker will truncate any data that doesn't fit in the type you're converting to.