Date: Tue, 14 Mar 2006 09:16:12 +0100 From: Mathieu Prevot <freebsd-amd64@club-internet.fr> To: Dan Nelson <dnelson@allantgroup.com> Cc: freebsd-hackers@freebsd.org, freebsd-amd64@freebsd.org Subject: Re: Creating real bool type for simulation in physics Message-ID: <20060314081612.GA55608@scienceclue.ath.cx> In-Reply-To: <20060313171341.GA7636@dan.emsphone.com> References: <20060313161740.GA56875@scienceclue.ath.cx> <20060313171341.GA7636@dan.emsphone.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Mar 13, 2006 at 11:13:41AM -0600, Dan Nelson wrote: > In the last episode (Mar 13), Mathieu Prevot said: > > I use freebsd/amd64 (RELENG_6) for simulation in physics. I am > > working on the Ising model: an assembly of spins (micromagnets) which > > interact and which are in one of two states (up or down). Until now I > > use char to define the state of each spin (-1 or 1), however, I > > remarked that most time is spent on memory I/O. Most of bits are > > unused. > > > > I think that if I can use just one bit per spin, I can have something > > much faster. I need advices on how to use it. I guess I can't define > > a new type with a 1/8 byte height (or one bit), yes ? What variable > > (int, char...) do you recommend to use for a sempron 64 bits. I think > > I'll need to define new operators (opaque operators, built with bit > > operators) to switch my spins or use directly the following: & | ^ ~ > > ... > > Take a look at the "bitstring" functions, which let you allocate an > array of "bits" and manipulate them individually. They're documented > in the bitstring manpage. Thank you. bitstring functions (macros) are based on char (8bits) type. This is not ANSI or POSIX, I will include /usr/src/sys/sys/bitstring.h in my program for portability. vi (nvi), and bind9 also use these macros, but are a bit different. -- Mathieu P http://scienceclue.ath.cx
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060314081612.GA55608>