Naomi
Elite
- Joined
- Jan 19, 2007
- Posts
- 4,005
- Society
- Rangers
- Avatar Name
- Naomi NP Polder
We live in the western world where bigger is supposed to mean better. Higher numbers equal GOOD. TV screen resolutions, cars HP, the size of apples and vegetables,...
This does not always work for coding, as far as I understand it. Please allow me to clarify:
(a 'click' means more or less the time memory needs to be accessed)
I'll use the following example:
A program gives 2 requests for 2x12 bits of memory code
Case 1: 32bit program, 32bit OS,
- click 1: 32bit consisting of 12bit code, + 12 bit code, + 8 bit unused;
- click 2: 32bit consisting of 12bit code, + 12 bit code, + 8 bit unused.
Case 2: 32bit program, 64bit OS, OS takes care of convergion,
- click 1: 64bit consisting of 2 blocks of 32bit code (consisting of 12bit code + 12bit code, + 8 bits unused)
Case 3: 64bit converged program, 64bit OS
- click 1: 64bit consisting of 12bit code, + 12 bit code, + 40 bit unused;
- click 2: 64bit consisting of 12bit code, + 12 bit code, + 40 bit unused.
Case 4: 64bit NATIVE program, 64bit OS:
- click 1: 64bit consisting of 4 blocks of 12bit code, + 16 unused (room for another request within this click
Conclusion: only Case 2 and Case 4 are fast, Case4 would have been even faster if the program needed 5 sets of 12bit code.
So... I wonder if EU 64-bit works so slow, because it works like Case 3 ?
feel free to assist when I'm wrong in my pov here
This does not always work for coding, as far as I understand it. Please allow me to clarify:
(a 'click' means more or less the time memory needs to be accessed)
I'll use the following example:
A program gives 2 requests for 2x12 bits of memory code
Case 1: 32bit program, 32bit OS,
- click 1: 32bit consisting of 12bit code, + 12 bit code, + 8 bit unused;
- click 2: 32bit consisting of 12bit code, + 12 bit code, + 8 bit unused.
Case 2: 32bit program, 64bit OS, OS takes care of convergion,
- click 1: 64bit consisting of 2 blocks of 32bit code (consisting of 12bit code + 12bit code, + 8 bits unused)
Case 3: 64bit converged program, 64bit OS
- click 1: 64bit consisting of 12bit code, + 12 bit code, + 40 bit unused;
- click 2: 64bit consisting of 12bit code, + 12 bit code, + 40 bit unused.
Case 4: 64bit NATIVE program, 64bit OS:
- click 1: 64bit consisting of 4 blocks of 12bit code, + 16 unused (room for another request within this click
Conclusion: only Case 2 and Case 4 are fast, Case4 would have been even faster if the program needed 5 sets of 12bit code.
So... I wonder if EU 64-bit works so slow, because it works like Case 3 ?
feel free to assist when I'm wrong in my pov here