Abstract: A data storage and retrieval system is provided which has extremely high capacity. The system includes a large array of small disk files, and three storage managers for controlling the allocation of data to the array, access to data, and the power status of disk files within the array. The allocation manager chooses the disk files upon which incoming data is written based on the current state of the disk files (active or inactive), the available capacity, and the type of protection desired (i.e unprotected, RAID), mirrored, etc.). The access manager interprets incoming read requests to determine the location of the stored data. The power manager sequences disk files between active and inactive to provide the storage requested by the access and allocation managers.

Abstract: An improved architecture for a Prolog interpreter/compiler is described to facilitate interrupt processing. The new architecture employs interrupt control words and a set of interrupt control blocks to enhance Prolog processing. As interrupts are detected, an interrupt control word is loaded with the address of an interrupt control block for managing interrupt processing. Then, each time a new predicate is fired, the interrupt control word is tested and control is passed to the interrupt processing routine if the control word is non-zero.

Abstract: An improved technique for processing Prolog objects is described. The first bit of the Prolog object word is used as a flag to indicate whether the object word is a type pointer or type descriptor. If the one-bit flag indicates that the object word is a type pointer, then the address is available for immediate processing without the unnecessary processing associated with the prior art.