Lsi Corporation

Lsi Logic Corporation

Vojislav Vukovic

Alexander Andreev

Ranko L Scepanovic

Sergey Vladimirovich Gribok

Igor A Vikhliantsev

Sergey Y Gribok

Ivan Pavisic

Ephrem C Wu

Nikola Radovanovic

Oleg N Izyumin

Brian Vanderwarn

A system for, and method of, assigning code blocks to constituent decoding units in a turbo decoding system having parallel decoding units. In one embodiment, the method includes: (1) representing the turbo decoding system as a resource diagram rectangle, (2) representing the code blocks as code block rectangles, (3) mapping the code block rectangles into the resource diagram rectangle and (4) assigning the code blocks to the constituent decoding units based on the mapping.

System and method for assigning code blocks to constituent decoder units in a turbo decoding system having parallel decoding units

Described embodiments provide a network processor that includes a security protocol processor to prevent replay attacks on the network processor. A memory stores security associations for anti-replay operations. A pre-fetch module retrieves an anti-replay window corresponding to a data stream of the network processor. The anti-replay window has a range of sequence numbers. When the network processor receives a data packet, the security hardware accelerator determines a value of the received sequence number with respect to minimum and maximum values of a sequence number range of the anti-replay window. Depending on the value, the data packet is either received or accepted. The anti-replay window might be updated to reflect the receipt of the most recent data packet.

Security protocol processing for anti-replay protection

A system for, and method of, assigning code blocks to constituent decoding units in a turbo decoding system having parallel decoding units. In one embodiment, the system includes: (1) a resource model generator configured to generate a model that represents the constituent decoding units and memories thereof along two dimensions, (2) a decoding unit number calculator associated with the resource model generator and configured to determine, for each of the code blocks, a number of the constituent decoding units to use to decode subblocks of each of the code blocks, (3) a rectangle mapper associated with the decoding unit number calculator and configured to generate a mapping in which the code blocks are mapped to the model and (4) a code block assigner associated with the rectangle mapper and configured to assign the subblocks of each code block to the constituent decoding units in accordance with the mapping.

Methods and apparatus are provided for programmable decoding of a plurality of code types. A method is provided for decoding data encoded using one of a plurality of code types, where each of the code types correspond to a communication standard. The code type associated with the data is identified and the data is allocated to a plurality of programmable parallel decoders. The programmable parallel decoders can be reconfigured to decode data encoded using each of the plurality of code types. A method is also provided for interleaving data among M parallel decoders using a communications network. An interleaver table is employed, wherein each entry in the interleaver table identifies one of the M parallel decoders as a target decoder and a target address of a communications network for interleaved data. Data is interleaved by writing the data to the target address of the communications network. The communications network can comprise, for example, a cross-bar switch and/or one or more first-in-first-out buffers.

Methods and apparatus for programmable decoding of a plurality of code types

An improvement to an arithmetic unit of a low-density parity-check decoder, where the arithmetic unit has a pipelined architecture of modules. A first module calculates a difference between absolute values of md_R and md_g_in, and passes the result to a first Gallager module. The first Gallager module converts this value from a p/prepresentation to a 2*p&#x2212;1 representation, and passes the result to a second module. The second module selectively adjusts the result of the previous module based on the sign values of md_g_in and md_R, and passes one of its outputs to a third module (the other two outputs, loc_item_out and hard_out, are not a part of the pipeline). The third module calculates a new md_g value by adding the result of the second module and loc_item_in, and passes this result to a fourth module. The fourth module separates a sign and an absolute value of the new md_g, and passes the result to a second Gallager module. The second Gallager module converts the result from the 2*p&#x2212;1 representation to the p/prepresentation and the final value leaves the unit as md_g_out. In these calculations, md_R=a check node value from the previous iteration, md_g=an edge value (md_g_in&#x2014;from the previous iteration, md_g_out&#x2014;for the next iteration), p=probability that a value is zero, p=probability that a value is one, loc_item_in/loc_item_out=intermediate values used for the md_g_out calculation, and hard_out=a bit value estimation for the current iteration of the pipelined arithmetic unit.

Pipelined LDPC arithmetic unit

A method for verifying the functionality of a repair system of configurable memory that functions to replace memory that fails predetermined tests with unused memory that passes the tests. The method includes the steps of providing a matrix comprising a plurality of reconfigurable memory blocks, providing an emulation system, generating a substitute memory block for each of the reconfigurable memory blocks utilizing the emulation system computing platform, providing a memory design that incorporates the substitute memory blocks, generating files for mapping errors into the reconfigurable memory blocks and providing a control file associated with the emulation system, and operating the emulation system to emulate the memory design.

RRAM memory error emulation

A frequency reduction or phase shifting circuit has an input receiving an input data stream having an input frequency and a representation of desired output frequency. A splitter splits the input data stream into a plurality of split signals each at a frequency of the desired output frequency. A plurality of catchers identify valid bits of each respective split signal. A shifter shifts valid bits identified by at least some of the catchers by a predetermined number, which establishes a de-serialization level for frequency reduction or phase shifting. An output provides an output data stream at the desired output frequency.

Data stream frequency reduction and/or phase shift

A decoder for access data stored in n memories comprises a function matrix containing addresses of the memory locations at unique coordinates. A decomposer sorts addresses from coordinate locations of first and second m&#xd7;n matrices, such that each row contains no more than one address from the same memory. Positional apparatus stores entries in third and fourth m&#xd7;n matrices identifying coordinates of addresses in the function matrix such that each entry in the third matrix is at coordinates that matches corresponding coordinates in the first matrix, and each entry in the fourth matrix is at coordinates that matches corresponding coordinates in the second matrix. The decoder is responsive to entries in the matrices for accessing data in parallel from the memories.

Decomposer for parallel turbo decoding, process and integrated circuit

A method for transforming a customer's memory design into an RRAM memory design. A port mapping table is created that lists the ports of the customer memories, and an instance types table is created that lists the customer memories. For each customer memory that is listed in the instance types table, any virtual buffer nets are removed, and any virtual buffers are removed. Any loose nets so created are reconnected to an RRAM cell in the RRAM memory design. The customer memory instance are then removed. A constraints file is updated from customer memory port designations to RRAM port designations. Automated test logic is inserted into the RRAM memory design, layout on the RRAM memory design is performed, and timing constraints on the RRAM memory design are satisfied. A modified version of the RRAM memory design is returned to the customer for verification. The modified version is made using the port mapping table. Each RRAM matrix is replaced with the customer memories it replaced, the removed virtual buffer nets and virtual buffers are left out, and other parts of the RRAM memory design are left unchanged.

RRAM backend flow

The present invention is directed to a spanning tree method for K dimensional space. To address timing driven buffer insertion and clock routing problems clusters of points must be constructed in 3-dimensional space. The first and second dimensions are coordinates on a plane, while the third dimension is time which is arrival pin time for buffers insertion and clock latency for clock routing. In a first aspect of the present invention, a method includes partitioning an input set of points into a binary tree of partitions so that each leaf partition has maximally a defined number of points. Graph edges are made for the points by connecting each point to its closest points in every of 2 subspaces and the number of graph nodes is then reduced to a predefined value.

Growing community of inventors

Santa Clara, CA, United States of America

Vojislav Vukovic