BoostCore

Provides the latest updates in the BoostCore documentation.

The Kunpeng Accelerator Engine (KAE) is a hardware-based acceleration solution on Kunpeng 920 series processors, including KAE encryption and decryption as well as KAE decompression.

The Kunpeng Unified Acceleration Framework (KUAF) is an in-house solution for enhancing the collaboration between hardware and software libraries for faster compression and encryption/decryption.

Snappy is a high-speed compression and decompression library. It provides extremely fast compression and decompression speeds rather than pursuing high compression ratios.

Zstandard (zstd) is a fast, lossless compression algorithm designed for real-time compression scenarios. It delivers fast compression and decompression while maintaining a competitive compression ratio.

LZ4 provides reasonable compression ratios. Its core advantages include extremely fast compression and decompression speeds, a small memory footprint, support for both streaming and block compression modes, and multi-language bindings.

zlib is a general-purpose DEFLATE compression library that provides support for zlib, gzip, and raw DEFLATE data formats.

The KACC_crypto library leverages CPU SIMD and pipelining to improve single-core computing efficiency, optimize the SHA1 MB algorithm, and enhance HMAC performance. It can connect to VPP applications.

KQMalloc is a memory allocator designed for Kunpeng 920 series processors. It has been adapted and optimized for Linux systems with 4 KB or 64 KB page sizes and 48-bit virtual memory addresses.

AVX2KI facilitates the porting of AVX intrinsics to the Kunpeng instruction set.

kpglibc leverages a vectorized instruction set to enhance the performance of glibc's string, memory, and time functions for Kunpeng processors.

KSL_ASN1 is optimized based on open-source ASN.1 software and has higher performance compared with asn1c. KSL_ASN1 supports the BER, DER, PER, and XER formats.

The Hyper Thread Library (HTL) is a user-level thread library built on kernel-mode threads. It solves application performance deterioration and system resource insufficiency when a large number of kernel-mode threads are used, especially when the performance is extremely low in nested parallel scenarios. This library improves concurrency and performance while reducing resource usage.

Kunpeng Vector Signal Processing (KVSIP) provides functions such as basic vector computing, basic matrix computing, and fast Fourier transform. These functions adapt to the Kunpeng platform.

Kunpeng Acceleration JSON, also referred to as KACC_JSON, is a high-performance JSON function library based on the Kunpeng platform. It uses SVE256 instructions to enhance JSON serialization and deserialization. By leveraging Kunpeng hardware advantages, KACC_JSON delivers higher performance than Gson, an open-source Java library.

This library runs on Kunpeng processors to offer acceleration features for the BiSheng JDK, including enhanced heap dump, JBooster, and JBolt.

The Homogeneous Acceleration Framework (HAF) provides easy-to-use APIs to effectively and securely offload service code to acceleration nodes for execution.

Hyper Media Performance Primitives (HMPP) provides high-performance acceleration functional interfaces for signal and image processing through NEON instructions supported by the Kunpeng processor.

By using hardware-accelerated vector instructions, KVCL enhances common basic operators for video encoding and improves their performance.

The Kunpeng Artificial Intelligence Library (KAIL) is a high-performance AI operator library optimized for the Kunpeng platform. It includes a deep neural network operator library (KDNN) and an extension operator library (KDNN_EXT).

Protobuf enables efficient data exchange and storage between distributed systems and supports seamless integration with RPC frameworks (such as gRPC).

fbthrift is a high-performance RPC framework and serialization library, widely used in distributed systems and microservice architectures. It supports multiple languages, and its serialization component provides a rich set of serialization protocols.

The performance of the serialization and deserialization functions in Sonic-Cpp is optimized using the vectorized instruction set based on Kunpeng processors. This library is applicable to Kunpeng 950 and the function interfaces are consistent with those of open-source Sonic-Cpp.

simdjson is a high-performance JSON parsing library. Through SIMD instruction set optimization, simdjson can parse JSON documents at a speed of several gigabytes per second, far outperforming traditional JSON parsers.

RapidJSON is a high-performance, lightweight, and header-only C++ JSON parser and generator. It is ideal for JSON data parsing, construction, validation, and serialization in server-side, embedded, and performance-sensitive scenarios.

The Kunpeng Math Library (KML) provides high-performance math interfaces optimized for the Kunpeng platform.

Smart prefetch is a storage acceleration library developed by Huawei. It uses high-speed cache drives and efficient prefetch algorithms to accelerate storage I/O, thus improving the overall performance in I/O-intensive scenarios.

As the virtual device layer, the Storage Performance Development Kit block device (SPDK bdev) interconnects with underlying virtual and physical devices. By enabling compression, encryption, and decryption in the bdev, all SPDK devices can support these features.

The Intelligent Storage Acceleration Library (ISA-L) is an open-source, high-performance storage algorithm library. On top of the open-source baseline, it leverages NEON and SVE instructions, as well as memory prefetch, to accelerate erasure coding and optimize cyclic redundancy checks (CRC).

This feature identifies the patterns of raw data and dynamically matches the optimal compression algorithm model to improve the data reduction rate. In addition, instruction-level optimization is performed for Kunpeng processors to accelerate compression.

Bcache is a block layer cache of the Linux kernel. It uses SSDs as the cache of HDDs for acceleration.

KSAL is a Huawei in-house storage algorithm library that uses algorithms optimized based on Kunpeng instead of mainstream open-source algorithms to improve storage performance.

KSML is a Huawei in-house storage maintenance library that provides fault prediction and slow drive detection.

This document describes how to use and tune the Data Plane Development Kit (DPDK) for network data packet acceleration on the Kunpeng processor.

Provides historical documents of Kunpeng BoostKit BoostCore.