64-bit TPT
-
HELP!!! it takes me to a page that says this is not the page you are looking for 404 error!
Edited once by cheeser101. Last: 7th Jan 2014 -
Why do you need a 64 bit verson of tpt? the ONLY advanteges with 64 bit is that you can use more than 4 gb of ram, which powder toy does not use more than, and sometimes if numbers get bigger than 4294967295 (2^32)-1 which is the highest number 32 bit computers can work with in straigt binary, if it goes any higher than the the cpu will have to do math by converting the numbers to binary coded decimal which is 4 bit "nibbles" that represent each digit, this is slightly slower, but there is basicly no advantage to 64 bit powder toy. Its is just misunderstandings conjoined with plucebo affect.
sorry to burst you bubble
-
Actually, with 64 bit there are more registers available, which can allow programs to run faster.
(However, pointers are larger, so 64 bit programs use more memory, which may reduce performance due to a greater number of cache misses. The only way to be sure is to do some benchmarks on the program of interest, 64 bit is not necessarily faster or slower than 32 bit in all situations.)
If you try to store a number greater than 2^32-1 in an unsigned 32 bit integer, it'll just overflow and wrap round to a smaller number. Binary coded decimal is not automatically involved, at least in C/C++ and therefore TPT. Pretty much everything in TPT works with 32 bit integers and floats anyway.
As I understand it, the normal way of doing arithmetic on 32 bit computers with numbers larger than 32 bits is to use several 32 bit numbers to store it, rather than BCD. http://x86asm.net/articles/working-with-big-numbers-using-x86-instructions/ has a few examples of assembly for addition and subtraction of 64 bit numbers on a 32 bit CPU. For longer numbers and more complicated operations like multiplication and division, some sort of external library is needed to do the maths, such as https://gmplib.org/. It isn't handled automatically by the CPU.
I believe the only reason you might ever do any calculations in BCD is to make it easy to display the result - but depending on how much arithmetic has to be done first, it might be faster to do it in binary then convert to BCD afterwards.
Edited 2 times by jacksonmj. Last: 7th Jan 2014 -
Interesting I did not know the part about the registers. And the BCD thing depends apoun the archeture of the cpu. Im studying cpu archetecture :D If I could download the mod I'd love to see how it benchmarks
-
What os are you using?
I can only compile a x64bit Linux or x86 Windows executable.
-
64 bit is a type of computer architechture (the way it's put together), that means that the processor can directly address any one of 2^65-1 memory nodues at any given time, meaning that it can process with optimum speed and efficiency. 32-bit machines are limited to using only 4 GB of RAM, however modern computers can have much more (up to 2^64 = ~ 18 447 Terabytes), meaning that processors can run much more intensive processes.
All architectures must use a power of 2 for the number of 'bits', meaning that you could theoretically have a 128-bit processor however this would be impractical as the adress space would greatly exceed the total amount of data stored on the Earth.
Edited 2 times by minecraft-physics. Last: 23rd Jan 2014
