Abstract: Systems and methods are provided for mitigating effects of hash collisions in hardware data compression, for example reducing or avoiding the side effects of hash collisions, or reducing or avoiding slow downs caused by hash collisions. In an aspect, a processor-implemented method includes: hashing an input data byte sequence to produce a hash value, the input data byte sequence being located at a sequence address within an input data stream; and storing, in a hash table at a hash address corresponding to the hash value, the sequence address and a portion of the input data byte sequence. In an aspect, to further avoid hash collisions, hash memory accesses are distributed among a plurality of parallel hash banks to increase the throughput. Another aspect virtually extends a hash depth by extending a data match search around broken hash links, going backward in the data sequence.

Abstract: Systems and methods are provided for mitigating effects of hash collisions in hardware data compression, for example reducing or avoiding the side effects of hash collisions, or reducing or avoiding slow downs caused by hash collisions. In an aspect, a processor-implemented method includes: hashing an input data byte sequence to produce a hash value, the input data byte sequence being located at a sequence address within an input data stream; and storing, in a hash table at a hash address corresponding to the hash value, the sequence address and a portion of the input data byte sequence. In an aspect, to further avoid hash collisions, hash memory accesses are distributed among a plurality of parallel hash banks to increase the throughput. Another aspect virtually extends a hash depth by extending a data match search around broken hash links, going backward in the data sequence.

Abstract: Systems and methods are provided that facilitate performing hardware acceleration processes without utilizing specialized drivers that are software and hardware specific by controlling the hardware accelerator with NVMe commands. The NVMe commands may be based on standardized NVMe commands provided in the NVMe specification, or may be vendor-specific commands that are supported by the NVMe specification. The commands are sent to the NVMe accelerator by a host CPU which, in some embodiments, may be located remotely to the NVMe accelerator. The NVMe accelerator may include a CMB on which a host CPU may set up an NVMe queue in order to reduce PCIe traffic on a PCIe bus connecting the CPU and the NVMe accelerator. The CMB may also be used by a host CPU to transfer data for acceleration to reduce bandwidth in the DMA controller or to remove host staging buffers and memory copies.

Abstract: Systems and methods are provided that facilitate performing hardware acceleration processes without utilizing specialized drivers that are software and hardware specific by controlling the hardware accelerator with NVMe commands. The NVMe commands may be based on standardized NVMe commands provided in the NVMe specification, or may be vendor-specific commands that are supported by the NVMe specification. The commands are sent to the NVMe accelerator by a host CPU which, in some embodiments, may be located remotely to the NVMe accelerator. The NVMe accelerator may include a CMB on which a host CPU may set up an NVMe queue in order to reduce PCIe traffic on a PCIe bus connecting the CPU and the NVMe accelerator. The CMB may also be used by a host CPU to transfer data for acceleration to reduce bandwidth in the DMA controller or to remove host staging buffers and memory copies.

Abstract: Systems and methods are provided for processing a non-volatile memory express over fabric (NVMe-oF) command at a Peripheral Component Interconnect Express (PCIe) attached accelerator device. Processing the NVMe-oF commands include receiving from a remote client, at a NVMe interface associated with the accelerator device, a Transport Control Protocol/Internet Protocol (TCP/IP)-encapsulated NVMe-oF command, and performing, at the accelerator device, functions associated with the NVMe-oF command that would otherwise be performed at a central processing unit (CPU).

Abstract: An FEC codeword comprises channel information indicating the reliability of the information contained by the FEC codeword. The channel information can be used to generate an initial error channel estimate. Based on the initial error channel estimate, an FEC decoder can decode the FEC codeword to increase the reliability of the information contained by the FEC codeword. According to the present disclosure, a method and system of decoding comprises: comparing a current codeword to a previous codeword in order to identify bits corrected between the previous and current codewords; revising an error channel estimate based on the identified corrected bits, the revised estimate representing a change in the error channel over time; and decoding the codeword based on the revised error channel estimate.

Abstract: A method and apparatus as described herein provide a novel modification to any iterative FEC decoder method that can improve FER performance in the error floor region. Many iterative FEC methods, such as commonly used LDPC decoders, have error floors where the performance of the decoder does not improve below a certain threshold. Error Floors are caused by trapping sets from which traditional methods cannot escape. With Stochastic Floor Mitigation, according to embodiments of the present disclosure, noise is strategically added to the operations occurring during decoding resulting in significantly improved error floor performance.

Abstract: A Forward Error Correction (FEC) decoder is provided, for example including a Layered Low Density Parity Check (LDPC) component. In an implementation, power consumption of the LDPC decoder is minimized with minimal to no impact on the error correction performance. This is achieved, in an implementation, by partially or fully eliminating redundant operations in the iterative process.

Abstract: A method and decoder are provided to decode a Low Density Parity Check codeword. An additional check processor performs hard-decision processing functions on the LDPC codeword in order to avoid running unnecessary decoder iterations. The method comprises: receiving the ECC codeword at a memory, the received ECC codeword comprising ECC data bits, ECC parity bits, and error detection code bits; soft-decision decoding the received ECC codeword at a soft-decision decoder, to update the ECC codeword according to ECC parity check equations; hard-decision processing the received ECC codeword at a check processor, while the soft-decision decoder performs the soft-decision decoding, to verify the ECC data bits using the error detection code bits; terminating the soft-decision decoding when the ECC data bits are verified, regardless of whether the updated ECC codeword satisfies all of the ECC parity check equations; and, outputting the decoded ECC codeword from the memory after termination of the decoding.

Abstract: A Forward Error Correction (FEC) decoder is provided, for example including a Layered Low Density Parity Check (LDPC) component. In an implementation, power consumption of the LDPC decoder is minimized with minimal to no impact on the error correction performance. This is achieved, in an implementation, by partially or fully eliminating redundant operations in the iterative process.

Abstract: A circuit for reducing phase errors in a digital communication systems signal is provided. The circuit comprises a demodulator block, a feed-forward path, a feed-back path, and a slicer. The demodulator block generates a plurality of samples from the signal and determines for each sample a corresponding phase error. The feed-forward path is configured to reduce in the signal a high frequency component of the phase errors. The feed-back path configured to reduce in the signal a low frequency component of the phase errors. The slicer selectively forwards phase errors to the feed-forward path or the feed-back path based on a respective magnitude of the phase error when operating in a decision-directed mode.

Abstract: In an aspect, in general, a forward error correction algorithm (FEC) utilizes an FEC block structure in a manner that extends the effective error correction such that it can approach an “infinite” length to obtain benefits typical of very large FEC block size without the commensurate computation cost.

Abstract: In an aspect, in general, a forward error correction algorithm (FEC) utilizes an FEC block structure in a manner that extends the effective error correction such that it can approach an “infinite” length to obtain benefits typical of very large FEC block size without the commensurate computation cost.

Abstract: An FFT processor using a single delay path and a permuter provides a reduction in the implementation area and a related reduction in power consumption through efficiencies obtained by the modification of a butterfly unit and the use of a novel interleaver. The modified butterfly unit is obtained by removal of complex variable multipliers, which is possible due to the simplification of twiddle factors in the stages that correspond to the modified butterfly unit.

Abstract: An interleaver for use with transform processors provides an address generator allowing for implementation using a reduced memory foot print, and permitting interleaving of an input sequence while minimizing latency.

Abstract: A single-path delay feedback pipelined fast Fourier transform processor comprising at least one set of triplet FFT stage means: a first FFT stage means comprising a radix-2 butterfly, a feedback memory, and a multiplication by unity; a second FFT stage means comprising a trivial coefficient pre-multiplication, a radix-2 butterfly, a feedback memory, and a multiplication by selectable unity or WNN/8; and a third FFT stage means comprising a trivial coefficient pre-multiplication, a butterfly, a feedback memory, and a complex twiddle coefficient multiplication with coefficients determined using a twiddle factor decomposition technique.