Abstract: Systems and techniques for advance cache allocation are described. A described technique includes selecting a job from a plurality of jobs; selecting a processor core from a plurality of processor cores to execute the selected job; receiving a message which describes future memory accesses that will be generated by the selected job; generating a memory burst request based on the message; performing the memory burst request to load data from a memory to at least a dedicated portion of a cache, the cache corresponding to the selected processor core; and starting the selected job on the selected processor core. The technique can include performing an action indicated by a send message to write one or more values from another dedicated portion of the cache to the memory.

Abstract: Systems and techniques for advance cache allocation are described. A described technique includes selecting a job from a plurality of jobs; selecting a processor core from a plurality of processor cores to execute the selected job; receiving a message which describes future memory accesses that will be generated by the selected job; generating a memory burst request based on the message; performing the memory burst request to load data from a memory to at least a dedicated portion of a cache, the cache corresponding to the selected processor core; and starting the selected job on the selected processor core. The technique can include performing an action indicated by a send message to write one or more values from another dedicated portion of the cache to the memory.

Abstract: A method and apparatus is provided for scheduling access to a common resource for a plurality of objects queued in a plurality of connection queues. Tokens associated with the connection queues are stored in scheduling queues. Each scheduling queue has a scheduling weight assigned thereto. Each connection queue has a connection weight value assigned thereto. A serving value is used to determine which scheduling queue to select. When a scheduling queue is selected, an object stored in a connection queue having an associated token stored in the selected scheduling queue is provided to the common resource. Tokens are moved among the scheduling queues as a function of the connection weight values, scheduling weights, and serving value. The objects queued in the connection queues may be fixed length cells or variable length packets.

Abstract: Disclosed herein is a system architecture capable of processing fixed length and/or variable length data packets. Under the method of the invention, incoming data packets are queued together according to their corresponding switch processing parameters (SPPs), and then the commonly-queued data packets are processed through a switch fabric as a single unit. In one aspect of the invention, the commonly-queued data packets are processed by the switch fabric as a single train packet. In another aspect of the invention, the commonly-queued data packets are sliced into a set of subtrain packets. A switch fabric then processes the set of subtrain packets in parallel using a plurality of switch planes. Both aspects of the invention can be implemented with a plurality of packet formatters and deformatters linked to a switch fabric in various configurations, including multi-path and hierarchical switching systems. a multichannel switching system.

Abstract: A method and apparatus is provided for scheduling access to a common resource for a plurality of objects queued in a plurality of connection queues. Tokens associated with the connection queues are stored in scheduling queues. Each scheduling queue has a scheduling weight assigned thereto. Each connection queue has a connection weight value assigned thereto. A serving value is used to determine which scheduling queue to select. When a scheduling queue is selected, an object stored in a connection queue having an associated token stored in the selected scheduling queue is provided to the common resource. Tokens are moved among the scheduling queues as a function of the connection weight values, scheduling weights, and serving value. The objects queued in the connection queues may be fixed length cells or variable length packets.

Abstract: Disclosed herein is a system architecture capable of processing fixed length and/or variable length data packets. Under the method of the invention, incoming data packets are queued together according to their corresponding switch processing parameters (SPPs), and then the commonly-queued data packets are processed through a switch fabric as a single unit. In one aspect of the invention, the commonly-queued data packets are processed by the switch fabric as a single train packet. In another aspect of the invention, the commonly-queued data packets are sliced into a set of subtrain packets. A switch fabric then processes the set of subtrain packets in parallel using a plurality of switch planes. Both aspects of the invention can be implemented with a plurality of packet formatters and deformatters linked to a switch fabric in various configurations, including multi-path and hierarchical switching systems. a multichannel switching system.