From ... From: Erik Naggum Subject: Re: more questions about threads... Date: 2000/04/05 Message-ID: <3163944702587593@naggum.no>#1/1 X-Deja-AN: 607159015 References: <8buulu$jp$1@nnrp1.deja.com> <38ea854a$0$21260@senator-bedfellow.mit.edu> <38EB2B22.B3AA3273@pindar.com> mail-copies-to: never Content-Type: text/plain; charset=us-ascii X-Complaints-To: newsmaster@eunet.no X-Trace: oslo-nntp.eunet.no 954956084 22113 195.0.192.66 (5 Apr 2000 17:34:44 GMT) Organization: Naggum Software; vox: +47 8800 8879; fax: +47 8800 8601; http://www.naggum.no User-Agent: Gnus/5.0803 (Gnus v5.8.3) Emacs/20.5 Mime-Version: 1.0 NNTP-Posting-Date: 5 Apr 2000 17:34:44 GMT Newsgroups: comp.lang.lisp * Tim Bradshaw | I'm not sure what if anything this says, except that solaris has fast | thread creation perhaps. I'd kind of expect their thread impl to be | pretty good because their market is big multiprocessor machines. I'd also expect Sun to want a "favorable demo" effect that shows this. Linux has made fork exceptionally fast (my experience is that PIDs are used up about three times faster under Linux than under SunOS 4.1.3_U1, with equivalent work being performed on the systems), and has implemented the performance sensitive parts of vfork in fork. ironically, it now appears that threads have to do _more_ work than processes because they _share_ the memory and a bunch of other stuff that fork splits off into objects with distinct identity. #:Erik