-------------------------------------------------------------------------------
				       
				       
		       B O S T O N   U N I V E R S I T Y
				       
			  Computer Science Department
				       
			      C O L L O Q U I U M
				       
			  Wednesday January 19, 1994
				       
				    3:00pm
			    (Tea served at 2:30pm)
				       
			    Seminar Room / MCS 135
				       

-------------------------------------------------------------------------------


			      THE SPRING KERNEL:

		 A Reflective, Distributed, and Multiprocessor
			  Real-Time Operating System

                                      by


			   Professor Jack Stankovic


			Department of Computer Science
			  University of Massachusetts
			    Amherst, Massachusetts


				       
				   ABSTRACT

	    Next  generation hard real-time systems require greater
	    flexibility and predictability  than is  commonly found
	    in today's systems.   These future systems include  the
	    space station, integrated vision/robotics/AI   systems,
	    collections of  humans/robots coordinating  to  achieve
	    common objectives  (usually in  hazardous  environments
	    such as undersea exploration  or chemical  plants), and
	    various command and  control applications.  The  Spring
	    kernel is a research oriented  kernel, designed to form
	    the   basis of   a flexible,  hard  real-time operating
	    system  for such applications.   The kernel advocates a
	    new   paradigm   for real-time   operating  systems and
	    attempts  to  provide   support  for  both   system and
	    application level predictability. The unique scheduling
	    approach found in the  kernel provides for an  on-line,
	    dynamic   guarantee of deadlines,   and the significant
	    amount of application level semantics acquired and used
	    by the kernel (its reflective nature) has the potential
	    to improve fault  tolerance and performance in overload
	    situations. Process management, memory management,  and
	    IPC all have unique  features with regard to supporting
	    predictability. The Spring kernel executes on a network
	    of multiprocessors (68030 and 68020 based) connected by
	    a predictable  fiber optic ring.  There exists a memory
	    hierarchy including a global replicated memory. In this
	    talk I   will discuss the  scalable, distributed memory
	    architecture,  the main  features  of  the kernel,  and
	    several   implementation  decisions      (based      on
	    predictability)  in each  of process management, memory
	    management including   support for  distributed  shared
            memory, and IPC.

            Host: Professor Azer Bestavros

-------------------------------------------------------------------------------
    For more information contact Prof. Azer Bestavros <bestavros@cs.bu.edu>
-------------------------------------------------------------------------------