Details of Grant 

EPSRC Reference:	GR/J84045/01
Title:	DYNAMIC USER-LEVEL OPERATING SYSTEM RESOURCE MANAGEMENT FOR PARALLEL LANGUAGES AND APPLICATIONS
Principal Investigator:	Warboys, Professor B
Other Investigators:	Gurd, Professor JR
Researcher Co-Investigators:
Project Partners:
Department:	Computer Science
Organisation:	Victoria University of Manchester, The
Scheme:	Standard Research (Pre-FEC)
Starts:	01 April 1994	Ends:	31 March 1997	Value (£):	159,448
EPSRC Research Topic Classifications:	Parallel Computing
EPSRC Industrial Sector Classifications:
Related Grants:
Panel History:
Summary on Grant Application Form
To produce a 'nanokernel' allowing user-level resource management. To investigate customised resource management for parallel applications. To provide an environment for customising and rescuing resource management code. To improve the performance of run-time systems for applications on parallel and distributed architectures.Progress:1. The hardware-dependent nanokernel, or hardware object, has been implemented for the Sparc processor of the EDS multicomputer. This successfully provides a platform for the user-level resource managers which have so far been implemented. Downcall interfaces for use by Process Manager, Persistent Store Manager, Store Manager and Network Manager are in place. Currently store management functions are provided directly by the hardware object. The upcall mechanism, whereby the nanokernel communicates with the user-level managers, is based on event handling at user-level and has been successfully implemented. At present the Sparc hardware object will support a single computation. Work is imminent to support multiple computations and multiple address spaces.2. A Process Manager class has been designed. The internal structure has been designed to be generic to allow easy customisation for particular policies. Several managers have been implemented, each providing a particular scheduling policy. The component of the Process Manager required for upcall event handling have been implemented. Event handling in Arena is designed to be extremely flexible and open to application-level policy. 3. A Persistent Store Manager class has been designed. A single-node implementation has been completed. Distributed persistent store management between RAMdisks on remote nodes is currently being designed and implemented . 4. Byte-stream I/O has been designed and has formed the basis for network management. A Network Manager has been designed and implemented, allowing work on distributed resource management. At present this includes persistent store management and terminal management. A Terminal Manager has been designed and implemented. This allows for canonical processing of user-input to threads of control executing C-library-level printf( ) and scanf( ) routines on remote nodes. 5. The system currently uses the GNU assembler, C and C++ compiler and linker tools, and the GNU C library has been modified to generate calls to manager and hardware object interfaces. Now that the basic Arena system design has been validated, work is imminent on modifying the linker to facilitate the dynamic linking of user-level resource managers. 6. Application code can be easily linked to the system, can be loaded onto several nodes, and can accept interactive user-input. The Arena source code is currently being moved into a version control system (GNU RCS) to facilitate work on the customisation environment.
Key Findings
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Potential use in non-academic contexts
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Impacts
Description This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk Summary Date Materialised
Sectors submitted by the Researcher
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Project URL:
Further Information:
Organisation Website: