floating point in the kernel
Hi, Can I use floating point operations inside the FreeBSD kernel version 4.7-stable? It used to be a policy not to use it for obvious performance probelms. Are there any known work arounds? Like some standard efficient techniques to convert floating point operations to fixed point operations without losing too much precision. Thanks, -Parveen. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: floating point in the kernel
Thanks for your response. My only concern is a square root operation I have to do. Is there some standard routine that will do it for long long or else I can pull out some code from gnu/i386/fpemul to do my computations. -Parveen. On Wed, 12 Feb 2003, Dan Nelson wrote: In the last episode (Feb 12), Parveen Patel said: Can I use floating point operations inside the FreeBSD kernel version 4.7-stable? It used to be a policy not to use it for obvious performance probelms. You're not supposed to. I think this was the last time it was brought up: http://groups.google.com/groups?threadm=746alr%24d44%241%40FreeBSD.csie.NCTU.edu.tw Are there any known work arounds? Like some standard efficient techniques to convert floating point operations to fixed point operations without losing too much precision. Multiply your numbers by 10^9, store everything in long longs, and write your own printf_fake_fp function that shifts the decimal point? -- Dan Nelson [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: floating point in the kernel
On Wed, Feb 12, 2003 at 07:50:23PM -0700, Parveen Patel wrote: Thanks for your response. My only concern is a square root operation I have to do. Is there some standard routine that will do it for long long or else I can pull out some code from gnu/i386/fpemul to do my computations. You probably should try to restructure your code so you don't need to do this. Kris msg19133/pgp0.pgp Description: PGP signature
Re: floating point in the kernel
In the last episode (Feb 12), Parveen Patel said: Can I use floating point operations inside the FreeBSD kernel version 4.7-stable? It used to be a policy not to use it for obvious performance probelms. You're not supposed to. I think this was the last time it was brought up: http://groups.google.com/groups?threadm=746alr%24d44%241%40FreeBSD.csie.NCTU.edu.tw Are there any known work arounds? Like some standard efficient techniques to convert floating point operations to fixed point operations without losing too much precision. Multiply your numbers by 10^9, store everything in long longs, and write your own printf_fake_fp function that shifts the decimal point? -- Dan Nelson [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: floating point in the kernel
In the last episode (Feb 12), Parveen Patel said: Thanks for your response. My only concern is a square root operation I have to do. Is there some standard routine that will do it for long long or else I can pull out some code from gnu/i386/fpemul to do my computations. Try http://www.azillionmonkeys.com/qed/sqroot.html , which has a whole bunch of sqrt functions. Also remember that if you're only comparing the sqrt to something else, it may be easier to square the rest of the equation instead. -- Dan Nelson [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message