spdlog is a high-performance, header-only or compiled C++ logging library. It provides synchronous and asynchronous logging to a wide variety of sinks including console (stdout/stderr with color), files (basic, rotating, daily, hourly), system facilities (syslog, systemd journal, Windows Event Log, Android), network (TCP, UDP, Kafka, Loki), and custom targets. The library uses the fmt library for formatting, supports custom pattern formatting with user-defined flags, backtrace logging, mapped diagnostic context (MDC), and configuration via environment variables or command-line arguments. It is designed for production use with extensive compile-time configuration options, thread safety, and minimal overhead. The architecture is modular: loggers manage a vector of sinks, each sink inherits from a base template that handles locking and formatting, and an optional async mode uses a thread pool with a bounded MPMC queue. The codebase is well-tested (Catch2) and benchmarked, with CI covering multiple compilers and platforms. It is a library, not an application, and is intended to be embedded into other C++ projects.
NVIDIA CUTLASS is a comprehensive C++ template library and Python DSL for implementing high-performance CUDA kernels, primarily focused on matrix multiply (GEMM), convolution, and attention operations. It provides a layered architecture from low-level warp matrix multiply-accumulate (MMA) and copy instructions (CuTe atoms) up through threadblock-level collective mainloops and epilogues, kernel-level schedulers (persistent, stream-K, warp-specialized), and a high-level operator API for runtime dispatch with fusion support. The repository includes a Python-embedded DSL (CuTeDSL) that uses MLIR-based compilation to generate optimized CUDA kernels, a separate Python code generation backend (cutlass_cppgen) for JIT compilation, and an extensive profiling and testing infrastructure. It spans NVIDIA GPU architectures from Volta to Blackwell and is actively maintained with contributions to deep learning frameworks like PyTorch and JAX.
Bazel is a multi-language, hermetic build system developed primarily by Google. The codebase serves as both the build tool itself (a C++ client communicating with a Java server) and the infrastructure for writing build rules in Starlark, a Python-like DSL. It implements a unique incremental build engine (Skyframe) that caches all intermediate states for correctness and speed, and supports remote caching and execution. The repository contains the full stack: command-line option parsing, Starlark interpreter and compiler, process sandboxing and wrapping, test runners (JUnit4 with coverage), jar merging, and platform-specific native code for Linux, macOS, and Windows. The architecture is split into a thin C++ client (handling startup, workspace discovery, and communication with the server) and a heavy Java server that performs all build analysis and execution. Key modules include the Starlark runtime (net.starlark.java.eval) which is an embeddable dynamic language with type checking, and the options framework (com.google.devtools.common.options) which provides annotation-driven, extensible command-line parsing. The repository is self-hosting: builds of Bazel are built with Bazel itself, using a bootstrap script. It maintains an extensive test suite across shell, Java, C++, and Python, and has active CI pipelines for multiple platforms and configurations.
AsmJit is a low-level C++ library for runtime machine-code generation and execution (JIT compilation). It is not an application but a reusable library, with its main entry points in `asmjit/core.h`, `asmjit/x86.h`, `asmjit/a64.h`, and `asmjit/ujit.h`. The library provides three emitter tiers built on a shared emitter abstraction: `BaseAssembler` for direct binary encoding, `BaseBuilder` for node-based IR manipulation, and `BaseCompiler` for high-level code with virtual registers, liveness analysis, and automatic register allocation. The x86/x64 and AArch64 backends are the primary production targets; AArch32 database infrastructure exists but is not a fully implemented backend.
JUCE is a mature, cross-platform C++ application framework specializing in audio software development. It provides a comprehensive set of modules covering core utilities, GUI, graphics, audio processing, MIDI, DSP, and plugin formats (VST3, AU, AAX, LV2, ARA). The architecture is a modular monolith: each module (e.g., juce_core, juce_audio_basics, juce_gui_basics) is a self-contained library that can be selectively included. Key entry points include JUCEApplication for desktop apps, AudioProcessor for audio plugins, and Component for GUI widgets. Data flows center on audio buffers (AudioSampleBuffer), MIDI messages, and event-driven GUI updates via message queues. External dependencies are minimal at the API level, but the framework bundles many third-party libraries (Opus, FLAC, Vorbis, WebP, HarfBuzz, Box2D, etc.) for codec and graphics support. The repository also includes the Projucer IDE for project generation and build tooling (CMake helpers, resource file management).
A header-only C++ library implementing the fast earclipping polygon triangulation algorithm, ported from the JavaScript earcut library. It provides a simple API (`mapbox::earcut`) that takes a polygon (with holes) and returns triangle indices, with an optional Delaunay refinement post-pass (`mapbox::refine`). The library is heavily optimized for real-time triangulation of vector tile geometries, using z-order curve hashing, custom memory allocation (object pool), and a block bounding-box index for efficient hole bridging. It targets C++14 for the public header and C++17 for tests/benchmarks. The repository also includes comprehensive test fixtures loaded at runtime from the upstream JS repository, benchmarks comparing against libtess2, and an interactive OpenGL visualizer.
ClickHouse is a high-performance, column-oriented OLAP database management system written primarily in C++. It is designed for real-time analytics on large datasets, supporting SQL queries, distributed execution, and various storage engines (MergeTree, ObjectStorage, etc.). The architecture is monolithic but modular, with distinct layers for parsing (Parsers), planning (Analyzer/Planner), execution (Processors), and storage (Storages). Key entry points include the server binary (`clickhouse-server`) and client tools (`clickhouse-client`). It relies heavily on C++ templates, RAII, and a plugin-like registration system for functions and storage engines. It supports distributed queries via ZooKeeper/Keeper and integrates with various cloud storage providers (S3, Azure, GCS).
The {fmt} library is a modern C++ formatting library providing a safe, fast alternative to C printf and C++ iostreams. It implements Python-like format string syntax (e.g., "{}", "{:d}", "{:.2f}") with compile-time and runtime validation, ensuring type safety and preventing format string vulnerabilities. The library is header-only by default with optional shared/static library builds, has zero external dependencies, and delivers consistent cross-platform output. Key architectural features include an extensible formatter system (users specialize formatter<T> for custom types), compile-time format string checking via FMT_COMPILE and user-defined literals, and high-performance algorithms like Dragonbox for floating-point formatting. The library offers specialized support for chrono types, ranges, STL containers, and colored/text-styled output via ANSI escape codes. Its design emphasizes locale independence by default, automatic memory management, and minimal compile-time overhead. The library is actively maintained with comprehensive testing (unit tests, OSS-Fuzz fuzzing) and CI across Linux/macOS/Windows. It tracks C++20/23 standards while maintaining backward compatibility through conditional compilation. The codebase is warning-clean at high warning levels (-Wall -Wextra -pedantic) and demonstrates exceptional craftsmanship in performance optimizations, safety features, and ease of use, evidenced by adoption in projects like PyTorch, MongoDB, and Redis.
The Solidity compiler (solc) is a production-grade, multi-pass compiler that translates Solidity smart contract source code into EVM bytecode. It supports two compilation pipelines: a legacy EVM assembly backend and a newer IR-based pipeline via Yul, which also serves as the optimizer intermediate representation. The optimizer suite comprises approximately 80 composable passes, including control-flow simplification, common subexpression elimination, function inlining, and store elimination. The compiler integrates formal verification through SMT solvers (Z3, cvc5, Eldarica) using both BMC and CHC engines. A C API (libsolc) enables embedding, and a CLI provides rich output options (ABI, bytecode, IR, metadata, storage layout). The codebase demonstrates deep compiler engineering with advanced static analysis, SSA construction, stack layout optimization, and genetic-algorithm-based optimizer tuning (yulPhaser). It is extensively tested with semantic, syntax, fuzz, and external integration tests.
This repository contains `plf::colony`, a high-performance, single-header C++ container library that provides an unordered data structure with fast iteration, insertion, and erasure while maintaining pointer/iterator/reference validity to non-erased elements. It is implemented as a template class in `plf_colony.h`, supported by `plf_tools.h` which provides compiler feature detection macros, utility functions, and allocator-aware helpers. The container uses a novel skipfield-based memory management technique (low-complexity jump-counting pattern) to achieve O(1) amortized insertion and erasure, and O(n) iteration with minimal overhead. It supports C++98 through C++23, with extensive compile-time feature detection and conditional compilation. The codebase is exceptionally well-crafted, demonstrating advanced template metaprogramming, low-level memory optimization, and careful handling of allocator traits, exception safety, and trivial type optimizations. The test suite (`plf_colony_test_suite.cpp`) is comprehensive, covering correctness, edge cases, fuzz testing, and exception recovery. The repository is a standalone library intended for reuse in any C++ project requiring a flexible, high-performance container with stable iterators.
This repository is the compiler toolchain for the Carbon Language, an experimental successor to C++ focused on performance, C++ interoperability, and memory safety. The codebase implements a full compilation pipeline from lexing and parsing through semantic analysis (SemIR) to LLVM IR code generation, along with a language server (LSP), a formatting tool, and C++ interop via Clang integration. The architecture follows a phased compiler design with distinct modules for lex, parse, check, sem_ir, lower, and driver, all orchestrated through a Bazel build system. Key entry points include the main driver binary (`toolchain/driver/`) with subcommands like `compile`, `link`, `build`, `format`, and `language-server`, and the semantic checker (`toolchain/check/`) which handles type checking, generics, and impl resolution. The toolchain also features a robust diagnostics system, a custom high-performance hashtable, and an extensive file-based testing framework with automatic CHECK-line updates. The codebase is actively developed with frequent commits, extensive test coverage, and fuzz testing, though it remains experimental and subject to breaking changes.
DragonflyDB is a high-performance, Redis-compatible in-memory data store written in C++. It implements the Redis wire protocol (RESP) and Memcached protocol, providing drop-in replacement capabilities with significantly higher throughput and lower latency, especially on multi-core machines. The architecture is based on a shared-nothing, multi-threaded design using fibers (boost.fibers) and a custom proactor event loop (epoll/io_uring). Data is partitioned across shards, each with its own thread, and transactions use a VLL-based locking scheme to minimize contention. The system supports replication via a write-ahead journal, cluster mode with slot migration, persistence via RDB snapshots, and tiered storage for offloading cold data to SSD. It also includes a full-text search engine with vector similarity (HNSW), JSON document store, Lua scripting, ACL-based security, and probabilistic data structures (Bloom, Cuckoo, Count-Min Sketch, Top-K). The codebase demonstrates advanced engineering in memory management (custom allocators, SIMD probing, compact object encoding) and concurrency (fiber-aware synchronization, RCU patterns). While primarily an application, it contains research-grade data structures and algorithms, giving it a hybrid character between production service and systems research project.
RocketRide is a high-performance data processing engine with modular nodes, AI capabilities, and cross-platform clients. The system provides a visual interface for building data pipelines that integrate various data sources, AI/ML services, and processing nodes. At its core, it features a C++-based execution engine for performance, Python-based processing nodes for flexibility, and multiple client applications (web, desktop, VS Code extension) for interaction. The platform emphasizes extensibility through its node system, allowing users to create custom processing components while providing built-in integrations for databases, cloud storage, AI services, and popular APIs. It includes monitoring, observability, and deployment capabilities for production use.
Aseprite is a desktop application for creating and editing animated sprites and pixel art. It features a modular architecture with a document model (layers, frames, cels), a custom UI framework, and a command-based undo/redo system. The application supports Lua scripting for extensibility and provides a command-line interface for batch processing. Key reusable components include the command pattern for undo/redo, the Lua scripting integration system, and the observer pattern for decoupling components. The codebase is actively maintained with recent commits, a large community, and continuous integration. Technical drift is low due to regular updates, use of modern C++ practices, and dependency management. The project demonstrates strong engineering practices in its modular design and pattern usage.
Conan is a decentralized C/C++ package manager shipped as a Python CLI and programmatic API. The code implements the complete package lifecycle: recipe authors define ConanFile classes, the CLI exports recipes, resolves dependency graphs from a local SQLite-backed cache and remote repositories, computes platform-specific package IDs, and installs prebuilt binaries or builds from source. It also contains a self-hosted package server (conans/server) with a Bottle-based REST API v2, authentication plugins, revision tracking, and file upload/download endpoints, though the primary product is the client tooling used by C/C++ developers and CI systems. Architecture is layered and modular. The CLI layer (conan/cli) uses a decorator-based command framework and delegates to a ConanAPI facade (conan/api) that orchestrates internal subsystems: graph resolution and lockfiles (conan/internal/graph), cache persistence (SQLite in conan/internal/cache/db), REST client and remote management (conan/internal/rest), recipe hooks and lifecycle methods (conan/internal/methods.py), and a broad set of build-system generators and toolchains under conan/tools covering CMake, Meson, MSBuild, Autotools, Bazel, Xcode, Qbs, Premake, SCons, and others. The main entry points are conans/conan.py:run for the CLI and conans/conan_server.py for the server. Data flows center on recipes and binary packages: export sources -> compute recipe revision, resolve requirement version ranges -> build dependency graph, compute package IDs from settings/options/conf, query local cache and remotes for binaries, then download/install or build/package as needed. Lockfiles pin revisions for reproducibility, profiles and per-package confs influence generated build-system files, and SBOM/audit features produce dependency metadata. The repository has an extensive three-tier test suite (unit, integration, functional) using in-process TestClient/TestServer harnesses and mocked network layers, plus platform-gated functional tests for real toolchains.
NCCL is NVIDIA's Collective Communications Library, a fundamental building block for multi-GPU and multi-node computing. It provides highly optimized implementations of standard collective operations (all-reduce, all-gather, reduce, broadcast, reduce-scatter) and point-to-point communication primitives. The library features a plugin architecture for extensibility, supporting various network transports (InfiniBand, sockets, shared memory), GPU Direct technologies, and remote memory access. With language bindings for Python and Rust, and extensive diagnostic and tuning capabilities, NCCL serves as the communication backbone for deep learning frameworks and HPC applications requiring efficient inter-GPU coordination.
Ladybird is a full-featured web browser engine implemented primarily in C++ with Rust components for performance-critical tasks. It includes its own JavaScript engine (LibJS) with bytecode interpreter and JIT compiler (Flap), a rendering engine (LibWeb) that implements HTML, CSS, DOM, SVG, WebGL, WebAudio, and many other web standards, a graphics library (LibGfx) with Skia integration, a media pipeline (LibMedia) supporting audio/video decoding via FFmpeg, and a custom garbage collector (LibGC). The architecture is multi-process: separate processes for UI, web content, image decoding, networking, and compositing, communicating via a custom IPC framework (LibIPC). The codebase is actively developed with a strong focus on web standards compliance, evidenced by extensive test suites including imported Web Platform Tests (WPT). It is not a data research project; it is a production-quality browser engine designed for the Ladybird browser, with cross-platform support for Linux, macOS, and Android.
Sunshine is a self-hosted, cross-platform game streaming server that implements the NVIDIA GameStream protocol, enabling remote play via Moonlight clients. It captures the host's display (using platform-specific backends such as DXGI, Wayland, X11, KMS, or PipeWire), encodes video with hardware acceleration (NVENC, VAAPI, Vulkan, AMF, QuickSync, or software), and forwards input (keyboard, mouse, gamepad, touch) from the client to the host. The server also provides a web-based configuration UI, handles client pairing with X.509 certificates and AES encryption, and manages streaming sessions via RTSP. It supports HDR, surround audio, gamepad feedback, and network QoS. Architecturally, Sunshine is a monolithic C++ application with a modular platform abstraction layer. The core components include an HTTP/HTTPS server for pairing and configuration, an RTSP server for session setup, and UDP broadcast threads for video, audio, and control data. The codebase is heavily optimized for low-latency streaming, featuring thread-safe queues, a generic task pool, and RAII resource management. It integrates with Boost, FFmpeg, libcurl, and various hardware encoder SDKs, and is packaged for Windows, Linux, macOS, and FreeBSD via multiple distribution channels. The repository demonstrates strong engineering practices with extensive unit and integration tests, consistent coding style, and active maintenance. While some files are large and platform-specific, the core utility modules are cleanly separated and reusable. The project is not a data/research repository; it is a production-grade application for game streaming enthusiasts and developers.
MLX is an array framework for machine learning on Apple Silicon and NVIDIA GPUs, developed by Apple's machine learning research team. It provides a NumPy-like API for Python and C++ with composable function transformations (autograd, vectorization, graph optimization), lazy computation, dynamic graph construction, and multi-backend support (CPU, Apple Metal, NVIDIA CUDA). The framework includes high-level neural network layers, optimizers, distributed training primitives, quantization support, and model I/O (safetensors, GGUF). It is designed for efficient on-device ML research and inference, with an emphasis on performance via custom GPU kernels, JIT compilation, and unified memory.
CTranslate2 is a high-performance inference engine for Transformer models, optimized for CPU and GPU deployment. It supports a wide range of architectures including encoder-decoder (BART, T5, NLLB, M2M-100, Whisper), decoder-only (LLaMA, GPT-2, Mistral, Gemma, Qwen), and encoder-only (BERT, DistilBERT, XLM-RoBERTa) models. The system provides a C++ library with Python bindings, a command-line tool (ct2-translator), and converters from popular frameworks like HuggingFace Transformers, Fairseq, Marian, OpenNMT-py/tf, and OPUS-MT. Key features include quantized inference (INT8/INT16/FP16/BF16/AWQ), flash attention, multi-GPU tensor parallelism, streaming generation, and token-level penalties. The architecture follows a modular design with device-agnostic operations, layer abstractions, and a model registry that allows plugging in new architectures without modifying the core engine. The codebase is actively maintained with frequent releases, extensive test coverage (C++ and Python), and CI pipelines for multiple platforms and backends.