Integrated Device Technology, Inc.

Mosys, Inc.

Advanced Micro Devices Corporation

Netlogic Microsystems, Inc

Intel Corporation

Monolithic Memories, Inc.

Step Engineering

Michael J Miller

Jay B Patel

Michael J Morrison

Mark W Baumann

John R Mick

Chuen-Der Lien

Richard Stephen Roy

Vladan Djakovic

Jeff D Smith

Kee W Park

Philip Arnold Ferolito

Scott Yu-Fan Chu

Philip A Bourekas

Dipak Sikdar

Bertan Tezcan

Bilal Murtaza

Robert J Proebsting

Chau-Chin Wu

John Raymond Mick, Jr

John M Birkner

Harmeet Bhugra

Andy Peng-Pui Chan

Avigdor Willenz

Darrell L Wilburn

Danh Le Ngoc

Warren K Miller

David Honig

Chih-Ping Sun

Robert W Stodieck

Chuen Der Lien

Chris Schott

Toru M Kuzuhara

Tuan Anh Luong

A system, method, and apparatus for graph memory. In one embodiment, the method includes: traversing program instructions disposed in an associative memory for operating a computer, the method comprising: receiving input data to be processed; identifying a next instruction to be fetched in the memory for processing the input data via: receiving a current node ID of a current state; performing a computational test on the input data resulting in a computed value; generating a search key by combining at least a portion of the computed edge value with the current node ID; and accessing the next instruction in associative memory via the search key.

Graph memory engine

A memory device that includes an input interface that receives instructions and input data on a first plurality of serial links. The memory device includes a memory block having a plurality of banks, wherein each of the banks has a plurality of memory cells, and wherein the memory block has multiple ports. An output interface provides data on a second plurality of serial links. A cache coupled to the IO interface and to the plurality of banks, stores write data designated for a given memory cell location when the given memory cell location is currently being accessed, thereby avoiding a collision. Memory device includes one or more memory access controllers (MACs) coupled to the memory block and one or more arithmetic logic units (ALUs) coupled to the MACs. The ALUs perform one or more operations on data prior to the data being transmitted out of the IC via the IO, such as read/modify/write or statistics or traffic management functions, thereby reducing congestion on the serial links and offloading appropriate operations from the host to the memory device.

Partitioned memory with shared memory resources and configurable functions

An apparatus, system, and method is described for calculating a composite index into a customizable hybrid address space that is at least partially compressed to locate a longest prefix match ('LPM') of a prefix string comprised of a plurality of multi-bit strides ('MBSs'). The device comprises: a mask-and-count logic for generating a base index into memory for a first MBS whose addresses are not compressed; a logical-shift apparatus that selectively uses a variable portion of the second MBS to generate an offset index from the given base index per an amount the second MBS addresses were actually compressed; and an add logic that adds the base index to the offset index to form the composite index that locates the LPM using a single access into memory. A compressed vector contains compression information of the second MBS in an information density format greater than a single bit to a single address.

System and method to match vectors using mask and count

System, method, and apparatus for integrated main memory (MM) and configurable coprocessor (CP) chip for processing subset of network functions. Chip supports external accesses to MM without additional latency from on-chip CP. On-chip memory scheduler resolves all bank conflicts and configurably load balances MM accesses. Instruction set and data on which the CP executes instructions are all disposed on-chip with no on-chip cache memory, thereby avoiding latency and coherency issues. Multiple independent and orthogonal threading domains used: a FIFO-based scheduling domain (SD) for the I/O; a multi-threaded processing domain for the CP. The CP is an array of independent, autonomous, unsequenced processing engines processing on-chip data tracked by SD of external CMD and reordered per FIFO CMD sequence before transmission. Paired I/O ports tied to unique global on-chip SD allow multiple external processors to slave chip and its resources independently and autonomously without scheduling between the external processors.

Integrated main memory and coprocessor with low latency

The randomizer includes connection circuitry with a random connection layout to parallely couple each of a quantity of bits for each of a plurality of inputs of bit width n to multiple output bits of a respectively coupled output. Combinational circuitry combines at least a portion of each of the plurality of outputs associated with each of the plurality of inputs to create a single resultant output of random data having a bit width of the quantity n.

Method and apparatus for randomizer

System, method, and apparatus for integrated main memory (MM) and configurable coprocessor (CP) chip for processing subset of network functions. Chip supports external accesses to MM without additional latency from on-chip CP. On-chip memory scheduler resolves all bank conflicts and configurably load balances MM accesses. Instruction set and data on which the CP executes instructions are all disposed on-chip with no on-chip cache memory, thereby avoiding latency and coherency issues. Multiple independent and orthogonal threading domains used: a FIFO-based scheduling domain (SD) for the I/O; a multi-threaded processing domain for the CP. The CP is an array of independent, autonomous, unsequenced processing engines that process on-chip data tracked by SD of external CMD and reordered per FIFO CMD sequence before transmission. Paired I/O ports tied to unique global on-chip SD allow multiple external processors to slave chip and its resources independently and autonomously without scheduling between the external processors.

A Method and Apparatus for Randomizer has been disclosed. In one implementation a sequence of randomized connections then exclusive-OR operations and then an adder produce a randomized output.

A memory device includes a block of memory cells and a cache. The block of memory cells is a random access memory with multiple ports. The block of memory cells is partitioned into subunits that have only a single port. The cache is coupled to the block of memory cells adapted to handle a plurality of accesses to a same subunit of memory cells without a conflict such that the memory appears to be a random access memory to said plurality of accesses. A method of operating the memory, and a memory with bank-conflict-resolution (BCR) module including cache are also provided.

Memory with bank-conflict-resolution (BCR) module including cache

A system includes a host device that may be configured to initiate memory requests to a system memory. The system also includes a memory controller that may be configured receive the memory requests and to format the memory requests into memory transactions that are conveyed to the memory device via a memory interface. The memory transactions include a memory write burst command having a first frame that includes a corresponding opcode positioned in one of a first command slot or a second command slot. The memory write burst command may also include a number of subsequent frames for conveying a data payload, as specified for example, by the opcode. The control unit may be configured to generate a number of concurrent sequential memory write operations to the memory in response to receiving the memory write burst command.

Memory system including variable write burst and broadcast command scheduling

A memory device that includes an input interface that receives instructions and input data on a first plurality of serial links. The instructions and input data are deserialized on the memory device, and are provided to a memory controller. The memory controller initiates accesses to a memory core in response to the received instructions. The memory core includes a plurality of memory partitions, which are accessed in a cyclic and overlapping manner. This allows each memory partition to operate at a slower frequency than the serial links, while properly servicing the received instructions. Accesses to the memory device are performed in a synchronous manner, wherein each access exhibits a known fixed latency.

High utilization multi-partitioned serial memory

A chip layout for a high speed semiconductor device is disclosed. The chip layout isolates Rx terminals and Rx ports from Tx terminals and Tx ports. A serial interface is centrally located to reduce latency, power and propagation delays. Stacked die that contain one or more devices with the chip layout are characterized by having improved latency, bandwidth, power consumption, and propagation delays.

Semiconductor chip layout

Semiconductor chip layout with staggered Tx and Tx data lines

A system for increasing the efficiency of data transfer through a serializer-deserializer (SerDes) link, and for reducing data latency caused by differences between arrival times of the data on the SerDes link and the system clock with which the device operates.

Reducing latency in serializer-deserializer links

An apparatus includes a receiver, an error detection unit, and an acknowledgement unit. The receiver may receive frames of data from a transmitter unit of a second apparatus via a first communication path. The error detection unit may detect data errors in the frames of data received via the first communication path. The acknowledgment unit may maintain an acknowledgement indicator indicative of whether frames received by the apparatus are error free. In response to the error detection unit detecting an error, the acknowledgement unit may indicate an error condition exists by freezing a value of the acknowledgement indicator, or alternatively the acknowledgement unit may set a current value of the acknowledgement indicator to a predetermined error value. Further, the apparatus may successively convey values of the acknowledgement indicator to the second apparatus via a second communication path while the apparatus is receiving frames.

Communication interface and protocol

Growing community of inventors

Saratoga, CA, United States of America

Michael J Miller