Pmc-sierra, Inc.

Pmc-sierra Us, Inc.

Microsemi Storage Solutions (u.s.), Inc.

Patrick Neil Bailey

Larrie Simon Carr

Heng Liao

Alexander John Cochran

A method of remapping allocated memory in a queue based switching element having first and second memory elements each allocated to a first port pair. An unallocated block of memory is identified in the first memory element as a candidate block, and an allocated block of memory is identified in the second memory element as a target block. Block information is copied from the target block to the candidate block, and the candidate block is maintained as unallocated. In response to a determination that read and write pointers are in a suitable position for a remapping operation, the candidate block is allocated and the target block is deallocated so that the second memory element becomes unallocated and available for reallocation to a second port pair.

Method for remapping of allocated memory in queue based switching elements

A method and apparatus are disclosed for management of linked lists within a dynamic queue system. In a dynamic queue system where a central memory is shared amongst a set of queues, the method organizes the linked list structures of the queues. The linked list pointers of the queues are organized over a set of single port memories. Memory for the queue entries is allocated in an alternating fashion, which allows the method to provide per-cycle access to queues while reducing the footprint of the memory elements used for maintaining the linked list structures. The method disclosed reduces the overall memory requirements for the design and implementation of queue systems with multiple queues sharing a common pool of memory.

Management of linked lists within a dynamic queue system

A method and apparatus for determining a set of common link rates for communication between two storage network elements in a storage network system. During the speed negotiation process, a controlling storage network element receives supported link rate information from a connected storage network element without providing any information in return. By not providing such information, although the speed negotiation process may not be completed, the controlling storage network element is still able to determine the supported link rates of the connected storage network element.

Method of rate snooping in a SAS/SATA environment

A desired FIFO buffer fill level is continuously derived during mapping or demapping of plesiosynchronous data signals into synchronized data signals, or vice-versa. One of j predefined integer values Iis repetitively consecutively produced during each consecutive one of j FIFO buffer write clock cycles, where i=1, . . . , j and where j and the integer values Iare selected such that closely approximates the number of bits read from the FIFO buffer per FIFO buffer write clock cycle. During each kconsecutive FIFO buffer write clock cycle, a Bits_Read value I+Iis produced where k=1, . . . , p; a Bits_Written value is produced; a Gap_Pattern value is derived by subtracting the Bits_Read value from the Bits_Written value; and, the Gap_Pattern is added to a predefined value representative a FIFO buffer center fill level to produce the desired FIFO buffer fill level.

Growing community of inventors

Coquitlam, Canada

Patrick Neil Bailey