Z From:	MERC::"uunet!CRVAX.SRI.COM!RELAY-INFO-VAX" 10-APR-1993 20:44 10-APR-1993 20:44:00.00 To:	Info-VAX@KL.SRI.COM  CC:	6 Subj:	Re: Interrupt Based Inter Process Communications  \ In article <1993Apr7.083819.371@gems.vcu.edu>, agnew@gems.vcu.edu (Brainwave Surfer) writes:+ > I never use mailboxes, too much overhead. I > Event flags and global common blocks, er, sections go like gangbusters.   M This "too much overhead" rumor about mailboxes is misleading.  It's true that O event flags (or hiber/wake) and global sections are generally faster; it's also * true that mailboxes are more convenient.    L On a ~6-VUP VAX (VS3100/76) a mailbox write (and read by the reader process)L takes a total of about a millisecond for a 1-K buffer.  (2k takes about 1.2 E msec)  Not terribly fast, but much faster than any real I/O device.     O Using a global section with either hiber/wake or common event flags takes about M a quarter of a second.  Definitely faster, but more work to set up.  I'll use N mailboxes whenever their attributes fit the application and where they're fastL enough -- which is most of the time.  Global sections seem to "fit"  better M in a "broadcast" sort of application, where one process updates a section and J many readers need to look at it.  Mailboxes fit better for message passing# from a writer to a single reader.     6 	--- Jamie Hanrahan, Kernel Mode Systems, San Diego CAO drivers, internals, networks, applications, and training for VMS and Windows NT N uucp 'g' protocol guru and release coordinator, VMSnet (DECUS uucp) W.G., and K Chair, Programming and Internals Working Group, U.S. DECUS VMS Systems SIG  B Internet:  jeh@cmkrnl.com  Uucp: uunet!cmkrnl!jeh  CIS: 74140,2055  