Dedication to the community
Süper Üye
64Bit is good but your computer needs more than 8gb of ram
64-bit integer instead of float in multiplayer games is a good idea?
- Konuyu başlatan CaioSilva00977
- Başlangıç tarihi
-
- Etiketler
- float float4 integer ue4 unreal engine unreal engine 4
- Durum
Onaylı Üye
if u have more than 8g ram u have to use 64bit any way
Süper Üye
use 64bit is better
Onaylı Üye
an integer uses alot of memory unlike string or char
Süper Üye
A float (32-bit) has a 23-bit mantissa, so at roughly 2^23 (about 8 million) it starts to not be able to represent each integer anymore. So if you assume that you want to place things in a game world at least at 1 millimeter precision, your largest game world that you can represent somewhat safely in floats is 8 kilometers along 1 dimension. That’s not bad actually, but a bit limiting.. and for precise physics 1mm may not be enough.
A double has 52 bit mantissa (2^52 == 4503599627370496), so that is about 4 billion kilometers at 1mm precision, and generally much higher precision for smaller worlds than that.
Still, doubles are inherently non-linear in their precision, so using 64-bit integers representing millimeter or an even smaller unit may be desirable for storing world data.
Of course, as Jeff hints at, you can also store things as 2 numbers, e.g. a 32-bit int for a world cell, and then a float to represent an offset inside that cell.
Then, when you’re ready to do some heavy calculations for physics or graphics, first translate all coordinates involved to be relative to the current cell, entity, or player, after which you can safely convert to float since all coordinates typically will be small.
A double has 52 bit mantissa (2^52 == 4503599627370496), so that is about 4 billion kilometers at 1mm precision, and generally much higher precision for smaller worlds than that.
Still, doubles are inherently non-linear in their precision, so using 64-bit integers representing millimeter or an even smaller unit may be desirable for storing world data.
Of course, as Jeff hints at, you can also store things as 2 numbers, e.g. a 32-bit int for a world cell, and then a float to represent an offset inside that cell.
Then, when you’re ready to do some heavy calculations for physics or graphics, first translate all coordinates involved to be relative to the current cell, entity, or player, after which you can safely convert to float since all coordinates typically will be small.
Üye
It's is good for performance in games
Onaylı Üye
Its faster then 32 bit programs...
Uzman Üye
but it also needs more specs
- Yasaklandı
- #36
Banlı Üye
64bit int will use a lot of memory
Onaylı Üye
depends on the specific locations, hardware,
Its right windows for games
Its right windows for games
Uzman Üye
64 bit always
- Durum