Abstract: A floorplan for a Structured ASIC chip is shown having a core region containing memory and VCLB logic cells surrounded by a plurality of IO connection fabrics that include a first IO connection fabric comprising IO sub-banks connecting the core of the chip to pins for external signals to the core, a first high-speed routing fabric disposed along the east-west vertical top of the core and connects the core to high-speed IO such as SerDes; a network-aware connection fabric connects the core to a microcontroller primarily for testing and repair of the memory in the core; and a second-high speed routing fabric is disposed on the north-south vertical sides of the core and communicates with the IO sub-banks. The VCLB Structured ASIC chip is manufactured on a 28 nm CMOS process lithographic node or smaller, having several metal layers and preferably is programmed on a single via layer.

Abstract: A parity unit circuit for use in a parallel, pipelined, low density parity check (LDPC) decoder that implements an iterative, min-sum, message passing LDPC algorithm. The parity unit provides a memory logic block for storing information relating to a current and next iteration of the LDPC computations and includes a “compute 1” logic block for computing a parity message (with sign) for application to related bit nodes and a “compute2” logic block for updating the data stored in the memory logic block for a next iteration of the LDPC decoder.

Abstract: 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.

Abstract: An LDPC decoder that implements an iterative message-passing algorithm, where the improvement includes a pipeline architecture such that the decoder accumulates results for row operations during column operations, such that additional time and memory are not required to store results from the row operations beyond that required for the column operations.

Abstract: A method of encoding a binary source message u, by calculating x:=Au, calculating y:=B?x, resolving the equation Dp=y for p, and incorporating u and p to produce an encoded binary message v, where A is a matrix formed only of permutation sub matrices, B? is a matrix formed only of circulant permutation sub matrices, and D is a matrix of the form D = ( T 0 … 0 0 0 T … 0 0 … … … … … 0 0 … T 0 I I … I I ) where T is a two-diagonal, circulant sub matrix, and I is an identity sub matrix.

Abstract: An LDPC decoder that implements an iterative message-passing algorithm, where the improvement includes a pipeline architecture such that the decoder accumulates results for row operations during column operations, such that additional time and memory are not required to store results from the row operations beyond that required for the column operations.

Abstract: A method of encoding a binary source message u, by calculating x:=Au, calculating y:=B?x, resolving the equation Dp=y for p, and incorporating u and p to produce an encoded binary message v, where A is a matrix formed only of permutation sub matrices, B? is a matrix formed only of circulant permutation sub matrices, and D is a matrix of the form D = ( T 0 … 0 0 0 T … 0 0 … … … … … 0 0 … T 0 I I … I I ) where T is a two-diagonal, circulant sub matrix, and I is an identity sub matrix.

Abstract: A parity unit circuit for use in a parallel, pipelined, low density parity check (LDPC) decoder that implements an iterative, min-sum, message passing LDPC algorithm. The parity unit provides a memory logic block for storing information relating to a current and next iteration of the LDPC computations and includes a “compute 1” logic block for computing a parity message (with sign) for application to related bit nodes and a “compute2” logic block for updating the data stored in the memory logic block for a next iteration of the LDPC decoder.

Abstract: A method of configuring a random access memory matrix containing partially configured memories in the matrix. The method includes the steps of independently calculating a memory enable signal and a configuration signal for a partially configured memory in each memory tile of the memory matrix. Memory tiles not supported by a memory compiler are determined. A memory wrapper is provided for each tile not supported by the memory compiler. An address controller is inserted in the memory matrix for each tile in a group of tiles. Output signals from each memory location in a memory group having a common group index are combined into a single output signal. A first stripe of memory tiles containing non-configured memory having a first width is selected. A second strip of memory tiles containing configured memory having a second width is also selected.

Abstract: Methods and apparatus are provided for a fast unbalanced pipeline architecture. A disclosed pipeline buffer comprises a plurality of memory registers connected in series, each of the plurality of memory registers, such as flip-flops, having an enable input and a clock input; and a controlling memory register having an output that drives the enable inputs of the plurality of memory registers, whereby a predefined binary value on an input of the controlling memory register shifts values of the plurality of memory registers on a next clock cycle. A plurality of the disclosed pipeline buffets can be configured in a multiple stage configuration. At least one of the plurality of memory registers can comprise a locking memory register that synchronizes the pipeline buffer. The pipeline buffer can optionally include a delay gate to delay a clock signal and an inverter to invert the delayed clock signal.

Abstract: 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.

Abstract: Methods and apparatus are provided for a fast unbalanced pipeline architecture. A disclosed pipeline buffer comprises a plurality of memory registers connected in series, each of the plurality of memory registers, such as flip-flops, having an enable input and a clock input; and a controlling memory register having an output that drives the enable inputs of the plurality of memory registers, whereby a predefined binary value on an input of the controlling memory register shifts values of the plurality of memory registers on a next clock cycle. A plurality of the disclosed pipeline buffets can be configured in a multiple stage configuration. At least one of the plurality of memory registers can comprise a locking memory register that synchronizes the pipeline buffer. The pipeline buffer can optionally include a delay gate to delay a clock signal and an inverter to invert the delayed clock signal.

Abstract: A method of encoding a binary source message u, by calculating x:=Au, calculating y:=B?x, resolving the equation Dp=y for p, and incorporating u and p to produce an encoded binary message v, where A is a matrix formed only of permutation sub matrices, B? is a matrix formed only of circulant permutation sub matrices, and D is a matrix of the form D = ( T 0 ? 0 0 0 T ? 0 0 ? ? ? ? ? 0 0 ? T 0 I I ? I I ) where T is a two-diagonal, circulant sub matrix, and I is an identity sub matrix.

Abstract: An LDPC decoder that implements an iterative message-passing algorithm, where the improvement includes a pipeline architecture such that the decoder accumulates results for row operations during column operations, such that additional time and memory are not required to store results from the row operations beyond that required for the column operations.

Abstract: A Gaussian noise is simulated by discrete analogue ri,j. A first parameter ? and pluralities of first and second integers i and j are selected. A plurality of points i,j are identified and a magnitude si,j is calculated for each point based on ?, i and j. The discrete analogue ri,j is based on a respective si,j. Examples are given of ? = 2 B - A 2 B and D>i?0 and 2C>j?0, where B?0, 2B>A>0, C?1 and D?1, and magnitude s i , j = 1 - ? i + ? i · 1 - ? 2 C · j ? ? ? or ? ? ? s D - 1 , j = 1 - ? D - 1 + ? D - 1 · 1 2 C · j . In some embodiments, a segment is defined based on ? and i. The segment is divided into points based on respective values of j, and the magnitude is calculated for each point of the segment. The defining and dividing segments and calculating the magnitude is iteratively repeated for each value of i.

Abstract: A method and apparatus are provided for creating and using a memory timing database. A plurality of characterization memories are defined, which can be mapped to a memory resource. Each characterization memory has different memory parameters. A plurality of variants of tiling each characterization memory to the memory resource are also defined. Timing characteristics of each tiling variant of each characterization memory are stored in the memory timing database for the memory resource based on sets of input ramptimes and output loads.

Abstract: A method of configuring a random access memory matrix containing partially configured memories in the matrix. The method includes the steps of independently calculating a memory enable signal and a configuration signal for a partially configured memory in each memory tile of the memory matrix. Memory tiles not supported by a memory compiler are determined. A memory wrapper is provided for each tile not supported by the memory compiler. An address controller is inserted in the memory matrix for each tile in a group of tiles. Output signals from each memory location in a memory group having a common group index are combined into a single output signal. A first stripe of memory tiles containing non-configured memory having a first width is selected. A second strip of memory tiles containing configured memory having a second width is also selected.

Abstract: A method of tiling a customer memory design to configurable memory blocks within a standardized memory matrix. A customer memory capacity and a customer memory width is determined for the customer memory design, and a standardized memory capacity and a standardized memory width is determined for the configurable memory blocks. The customer memory capacity and the customer memory width are selectively transformed by inverse factors based at least in part on a comparison of the customer memory capacity and the standardized memory capacity. Case independent blocks are formed within the configurable memory blocks, where the case independent blocks include gate structures formed in a standardized array in a substrate in which the customer memory design is to be implemented.

Abstract: A memory generation and placement flow system that receives a customer memory design and places the customer memory design within a customizable standardized integrated circuit design. The memory generation and placement flow system includes a memory librarian tool, a memory estimator tool, and a memory placer tool.

Abstract: Apparatus and process identifies a maximum or minimum value among a plurality of binary values on a plurality of a-bit wide wires in an integrated circuit module. An N-bit vector K is calculated based on n most significant bits of all a-bit binary signals, where N=2n. M N-bit vectors K—0, . . . ,K_(M?1) are calculated based on the n most significant and the m least significant bits of all a-bit binary signals, where M is at least 2m?1. A table is constructed from vectors K—0, . . . ,K(M?1) to create table vectors. A table vector is selected based on vector K, is used to derive a vector P, which in turn is used to select another table vector. The minimum or maximum binary value is identified from the two selected table vectors.