Author: Vincent Diepeveen
Date: 16:17:43 01/19/05
Go up one level in this thread
On January 19, 2005 at 15:17:15, Dann Corbit wrote: Practical working with 'int' IS the fastest way to work with generic code at modern processors. I did extensive 8 bits versus 'unsigned int' versus 'int' comparisions and 'int' just is either faster or equally fast than anything else at modern processors. What most write here about 'unsigned int' being 5% faster is major bullshit to say it polite. What happens is that they are everywhere type mixing in their code and that 'unsigned int' is less likely to give partial register stalls. However, signed 64 bits integers at opteron i cannod advice. >On January 19, 2005 at 11:55:43, Rémi Coulom wrote: > >>Hi, >> >>I am porting my chess program to 64 bit on an AMD64 Athlon processor. I have >>noticed that, with gcc, sizeof(int)=4. I would have expected sizeof(int)=8. On >>32-bit platforms, 32-bit variables are faster than 16-bit variables. I wonder if >>64-bit variables are faster than 32-bit variables on 64-bit machines. I have >>made a few experiments and did not notice much difference. > >It was the original intention for the C language to use int as the fastest >integer type. That has gone by the wayside for some reason. However, there are >new hints in the C99 spec for fast types that some compilers should support. > >From the ANSI/ISO C99 Standard: > >7.18.1.3 Fastest minimum-width integer types >1 Each of the following types designates an integer type that is usually fastest >2 The typedef name int_fastN_t designates the fastest signed integer type with a >width of at least N. The typedef name uint_fastN_t designates the fastest >unsigned integer type with a width of at least N. [see: footnote 216] >3 The following types are required: >int_fast8_t >int_fast16_t >int_fast32_t >int_fast64_t >uint_fast8_t >uint_fast16_t >uint_fast32_t >uint_fast64_t >All other types of this form are optional. > >footnote 216: The designated type is not guaranteed to be fastest for all >purposes; if the implementation has no clear grounds for choosing one type over >another, it will simply pick some integer type satisfying the signedness and >width requirements.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.