Topics / #high-performance
High-performance projects on GitHub
Software engineered for maximum speed and low latency
As of , the most-starred High-performance projects on GitHub are llama.cpp (131.4K stars), tauri (112K stars) and ripgrep (69.2K stars). GitHub.Locker tracks 141 High-performance repositories with 2.6M stars combined. The fastest-growing on October 16 was BRUTAL, with +340 stars in a day.
Hot right now
Most stars gained in a day · Oct 16
+340BRUTAL
ahwootech/BRUTAL · ★ 1.6KBRUTAL is a free, open source C# framework for building games or your own game engine, giving C# direct access to native APIs and Vulkan plus the low-level libraries around them (memory, maths, collections, job scheduling, and more).
+124llama.cpp
ggml-org/llama.cpp · ★ 131.4Kllama.cpp is a plain C/C++ implementation for running LLM (and VLM) inference with minimal setup and state-of-the-art performance, either locally or in the cloud.
FalkorDB
FalkorDB/FalkorDB · ★ 10.3KFalkorDB is an ultra-fast graph database built on sparse matrix math (GraphBLAS) instead of traditional traversal, designed to power Knowledge Graphs for LLM applications like GraphRAG, agent memory, and fraud detection.
+59ComfyUI-Ref2VA-VSA
Kablex/ComfyUI-Ref2VA-VSA · ★ 570It's a ComfyUI custom node that dramatically speeds up Reference-to-Video (Ref2VA) generation for the MiniMax H3 model, producing a full 5-second, 24fps character video in about 72 seconds on a single RTX 4090 instead of over 10 minutes.
The all-time top
Top 100 of 141 · by stars
- 131.4K#1 · ggml-org/llama.cpp
llama.cpp
llama.cpp is a plain C/C++ implementation for running LLM (and VLM) inference with minimal setup and state-of-the-art performance, either locally or in the cloud.
- 112K
#2 · tauri-apps/tauritauri
Tauri is a framework for building tiny, blazingly fast binaries for all major operating systems, letting you write the UI with any front-end framework that compiles to HTML, JS and CSS.
- 69.2K
#3 · BurntSushi/ripgrepripgrep
ripgrep (rg) is a line-oriented command-line tool that recursively searches directories for a regex pattern, automatically respecting .gitignore rules and skipping hidden and binary files by default.
- 4
alacritty
66Kalacritty/alacritty · RustAlacritty is a modern, cross-platform terminal emulator that renders via OpenGL and comes with sensible defaults while still allowing extensive configuration.
- 5
minio
61.3Kminio/minio · GoMinIO is a high-performance, S3-compatible object storage server, built to power AI/ML, analytics, and other data-intensive workloads.
- 6
julia
49.2KJuliaLang/julia · JuliaJulia is a high-level, high-performance dynamic programming language built for technical and scientific computing, numerical analysis, and machine learning.
- 7
hyperswitch
45.3Kjuspay/hyperswitch · RustHyperswitch is an open-source, composable payments platform that lets businesses route, retry, vault, and reconcile payments across 120+ payment processors like Stripe, Adyen, Braintree, and Worldpay, either self-hosted or as a SaaS.
- 8
faiss
41.2Kfacebookresearch/faiss · C++Faiss is a library for efficient similarity search and clustering of dense vectors, developed primarily at Meta's Fundamental AI Research group. It can search collections of vectors of any size, even ones that do not fit in RAM.
- 9
tidb
40.6Kpingcap/tidb · GoTiDB is an open-source, cloud-native distributed SQL database written in Go — MySQL-compatible and built for high availability, horizontal scalability, strong consistency, and mixing transactions with analytics.
- 10
fiber
40.2Kgofiber/fiber · GoFiber is an Express-inspired web framework for Go, built on top of the fasthttp engine, with fast development and high performance in mind.
- 11
polars
40.1Kpola-rs/polars · RustPolars is an extremely fast analytical query engine for DataFrames, written in Rust and designed to be fast, easy to use and expressive.
- 12
servo
38.2Kservo/servo · RustServo is a prototype web browser engine written in Rust. The project aims to give developers a lightweight, high-performance alternative for embedding web technologies in applications.
- 13
interview_internal_reference
37.3K0voice/interview_internal_reference · PythonA curated collection of real technical interview questions and answers from major Chinese tech companies — Alibaba, Tencent, Baidu, Meituan, ByteDance and more — with analysis from expert interviewers.
- 14
rustfs
35Krustfs/rustfs · RustAn open-source, high-performance distributed object storage system written in Rust. It speaks the S3 API, is optimized for data lakes, AI and big data workloads, and supports migration and coexistence with MinIO and Ceph.
- 15
kafka
34Kapache/kafka · JavaApache Kafka is an open-source distributed event streaming platform used by thousands of companies for high-performance data pipelines, streaming analytics, data integration, and mission-critical applications.
- 16
tokio
33.4Ktokio-rs/tokio · RustTokio is a runtime for writing reliable, asynchronous applications with Rust: it supplies the scheduling, non-blocking I/O, networking and timers that async Rust programs need.
- 17
go-zero
33.4Kzeromicro/go-zero · Gogo-zero is a cloud-native Go framework for building web and RPC microservices, packed with built-in engineering practices. It has been serving sites with tens of millions of users for years and is listed in the CNCF Landscape.
- 18
echo
32.8Klabstack/echo · GoEcho is a high performance, minimalist web framework for Go, used to build web apps, RESTful APIs and microservices.
- 19
beego
32.4Kbeego/beego · GoBeego is an open-source, high-performance web framework for Go, aimed at rapid development of enterprise applications: RESTful APIs, web apps and backend services.
- 20
rocksdb
32.2Kfacebook/rocksdb · C++RocksDB is an embeddable, persistent key-value store for fast storage, built as a C++ library and developed by Facebook's Database Engineering team.
- 21
dragonfly
31.8Kdragonflydb/dragonfly · C++An in-memory data store built as a modern replacement for Redis and Memcached — fully compatible with both APIs, so no code changes needed; it claims 25X more throughput on up to 80% fewer resources.
- 22
turborepo
31.2Kvercel/turborepo · RustTurborepo is a high-performance build system for JavaScript and TypeScript codebases (especially monorepos), written in Rust — its README now introduces it as 'the build system for coding agents'.
- 23
folly
30.6Kfacebook/folly · C++Folly (Facebook Open Source Library) is a library of C++20 components built for practicality and efficiency, used extensively at Facebook and often a dependency of its other open-source C++ projects.
- 24
srs
29.3Kossrs/srs · C++SRS (Simple Realtime Server) is a simple, high-performance, AI-driven real-time media server for live streaming: RTMP, WebRTC, HLS, HTTP-FLV, SRT, MPEG-DASH, and GB28181, with codecs from H.264 to AV1, VP9, AAC, Opus, and G.711.
- 25
xgboost
28.9Kdmlc/xgboost · C++XGBoost is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable — it implements machine learning algorithms under the Gradient Boosting framework.
- 26
druid
28.2Kalibaba/druid · JavaDruid is an open-source, high-performance Java database connection pool and SQL parser built by Alibaba's DataWorks team. It combines a JDBC connection pool, a multi-dialect SQL parser/AST analyzer, a SQL firewall, and monitoring statistics into one library.
- 27
zstd
28Kfacebook/zstd · CZstandard (zstd) is a fast lossless compression algorithm for real-time scenarios, at zlib-level ratios or better. This repo is the reference C implementation plus a CLI; the format is stable and documented in RFC8878.
- 28
pingora
27.6Kcloudflare/pingora · RustPingora is a Rust framework by Cloudflare for building fast, reliable, and programmable network services and proxies. It is battle-tested, having served more than 40 million internet requests per second for years.
- 29
valkey
27.4Kvalkey-io/valkey · CValkey is an open-source, high-performance data structure server for key/value workloads like caching and other realtime use cases. It ships a wide range of native structures plus a plugin system for adding new ones.
- 30
elixir
26.7Kelixir-lang/elixir · ElixirElixir is a programming language designed for building scalable and maintainable applications, running on the Erlang VM (BEAM).
- 31
miaosha
26.6Kqiurunze123/miaosha · JavaThis is a Chinese learning repository about designing and building a high-concurrency 'flash sale' (seckill) system for internet engineers, combining iterative demo projects, small frameworks, study notes and diagrams covering distributed-systems topics.
- 32
simdjson
24.4Ksimdjson/simdjson · C++A C++ library that parses JSON at gigabytes-per-second speed using SIMD instructions and microparallel algorithms. It ships in production systems like Node.js, ClickHouse and Meta Velox.
- 33
Paddle
24.1KPaddlePaddle/Paddle · C++PaddlePaddle is an open-source deep learning platform — China's first independently developed one, open-sourced since 2016 — spanning a core framework, model libraries, end-to-end dev kits and service platforms.
- 34
timescaledb
23.7Ktimescale/timescaledb · CTimescaleDB is a PostgreSQL extension for high-performance real-time analytics on time-series and event data — sensor readings, financial ticks, application events. You work with it in regular SQL.
- 35
fasthttp
23.5Kvalyala/fasthttp · GoFastHTTP is a fast HTTP server and client library for Go, tuned for high performance: zero memory allocations in hot paths and up to 10x the speed of net/http.
- 36
chi
23Kgo-chi/chi · Gochi is a lightweight, idiomatic and composable router for building Go HTTP services, aimed especially at large REST APIs that stay maintainable as the project grows and changes.
- 37
vector
22.7Kvectordotdev/vector · RustAn open-source, high-performance observability data pipeline: it collects, transforms, and routes your logs and metrics to any vendors you choose, so you keep control of your observability data. It is maintained by Datadog's Community Open Source Engineering team.
- 38
rocketmq
22.6Kapache/rocketmq · JavaApache RocketMQ is an open-source distributed messaging and streaming platform for building event-driven applications, with low latency, trillion-level capacity and flexible scalability.
- 39
openobserve
22.4Kopenobserve/openobserve · TypeScriptOpenObserve is an open-source, cloud-native observability platform for logs, metrics, traces, frontend monitoring (RUM), session replay and AI/LLM observability — a cost-effective alternative to Datadog, Splunk and Elasticsearch.
- 40
go-redis
22.3Kredis/go-redis · GoThe official Redis client library for Go: you import it into your Go program to talk to Redis servers. It targets the last three Redis releases and speaks both RESP2 and RESP3.
- 41
onnxruntime
22.1Kmicrosoft/onnxruntime · C++ONNX Runtime is a cross-platform, high performance ML accelerator for inference and training. It runs models from PyTorch, TensorFlow/Keras and classical ML libraries like scikit-learn, LightGBM and XGBoost.
- 42
dgraph
21.8Kdgraph-io/dgraph · GoDgraph is an open-source, horizontally scalable distributed graph database with a GraphQL interface, built for real-time queries over terabytes of data. It provides ACID transactions, consistent replication, and linearizable reads.
- 43
HikariCP
21.2Kbrettwooldridge/HikariCP · JavaA high-performance JDBC connection pool for Java: it keeps a ready pool of database connections your app can reuse instead of opening a new one for every request. It is billed as zero-overhead and production ready.
- 44
nats-server
20.9Knats-io/nats-server · GoThis is the NATS server — the high-performance core of NATS.io, a messaging system for connecting digital systems, services and devices. It's a CNCF project.
- 45
triton
20.4Ktriton-lang/triton · MLIRTriton is a language and compiler for writing highly efficient custom Deep-Learning GPU kernels, aiming to give higher productivity than CUDA and more flexibility than other existing DSLs.
- 46
ip2region
19.7Klionsoul2014/ip2region · GoAn offline IP-to-region localization library and data management framework: it maps IPv4 and IPv6 addresses to city-level regions with 10-microsecond queries, shipping xdb data files plus query clients for many programming languages.
- 47
ktransformers
19.6Kkvcache-ai/ktransformers · PythonKTransformers is a research framework for efficient inference and fine-tuning of large language models through CPU-GPU heterogeneous computing, letting huge MoE models run on hardware that doesn't have enough VRAM to hold them entirely on GPU.
- 48
uWebSockets
19KuNetworking/uWebSockets · C++μWebSockets is a simple, secure and standards-compliant web server for demanding applications, built as a C++ library — so optimized for speed and memory that its TLS 1.3 encrypted messaging outruns most servers' cleartext.
- 49
swoole-src
18.9Kswoole/swoole-src · C++Swoole is an event-driven, asynchronous, coroutine-based concurrency library for PHP, built for high performance.
- 50
disruptor
18.5KLMAX-Exchange/disruptor · JavaLMAX Disruptor is a Java library for high-performance messaging between threads inside a single application — very fast in-process event passing.
- 51
Megatron-LM
18.2KNVIDIA/Megatron-LM · PythonNVIDIA's open-source toolkit for training large transformer (LLM) models at scale. It comes as Megatron-LM, a reference setup with ready training scripts, and Megatron Core, a composable library of GPU-optimized building blocks for training models across thousands of GPUs.
- 52
caffeine
17.9Kben-manes/caffeine · JavaCaffeine is a high-performance, near-optimal in-memory caching library for Java with a Guava-inspired API.
- 53
VictoriaMetrics
17.9KVictoriaMetrics/VictoriaMetrics · GoVictoriaMetrics is a fast, cost-effective monitoring solution and time series database. It stores metrics long-term and can serve as storage for Prometheus or a drop-in replacement for Prometheus and Graphite in Grafana.
- 54
mars
17.7KTencent/mars · C++Mars is a cross-platform network component library built by WeChat's mobile team — the same infrastructure proven by billions of WeChat users. It bundles mobile-optimized networking, logging and network diagnostics for embedding in your own app.
- 55
brpc
17.6Kapache/brpc · C++bRPC is an industrial-grade C++ RPC framework built at Baidu for high-performance systems like search, storage, machine learning, advertisement and recommendation; the name stands for 'better RPC'.
- 56
instant-ngp
17.6KNVlabs/instant-ngp · CudaNVIDIA Research's implementation of four neural graphics primitives — neural radiance fields (NeRF), signed distance functions, neural images and neural volumes — fast enough to train a fox NeRF in under 5 seconds.
- 57
questdb
17.4Kquestdb/questdb · JavaQuestDB is an open-source, low-latency time-series database on open formats: it ingests millions of events per second, computes on live data in milliseconds, and keeps years of history queryable through one SQL engine.
- 58
neo4j
17.3Kneo4j/neo4j · JavaNeo4j is the world's leading graph database: a high-performance store where data is modeled as a flexible network of nodes and relationships instead of static tables, with a friendly query language and ACID transactions.
- 59
tigerbeetle
17.2Ktigerbeetle/tigerbeetle · ZigTigerBeetle is a financial transactions database built for mission-critical safety and performance, designed to power OLTP (online transaction processing) workloads for decades to come using debit/credit transfer primitives.
- 60
ceph
17.1Kceph/ceph · C++Ceph is an open-source distributed storage platform: it pools a cluster of servers into one scalable storage system that serves object, block, and file storage at the same time.
- 61
kcp
16.9Kskywind3000/kcp · CKCP is a fast, reliable ARQ protocol for realtime traffic: it trades 10-20% extra bandwidth for 30-40% lower average latency and up to 3x lower maximum latency than TCP.
- 62
codon
16.9Kexaloop/codon · PythonCodon is a high-performance Python implementation that compiles your code straight to native machine code — typically 10-100x faster than vanilla Python on a single thread, on par with C/C++.
- 63
muduo
16.3Kchenshuo/muduo · C++Muduo is a multithreaded C++ network library based on the reactor pattern, built for writing high-performance, event-driven servers on Linux.
- 64
scylladb
15.8Kscylladb/scylladb · C++Scylla is a real-time big data database — a distributed NoSQL data store that is API-compatible with Apache Cassandra and Amazon DynamoDB and built on the Seastar framework.
- 65
gjson
15.6Ktidwall/gjson · GoGJSON is a Go library for pulling values out of JSON documents quickly — one call with a path like 'name.last' returns the value. It also supports iteration and JSON Lines.
- 66
gpu.js
15.5Kgpujs/gpu.js · JavaScriptGPU.js is a JavaScript library that takes the functions you write and runs them on your GPU as shaders instead of on the CPU — a big speedup for parallel math like matrix multiplication.
- 67
TensorRT-LLM
14.8KNVIDIA/TensorRT-LLM · PythonTensorRT LLM is NVIDIA's Python framework for defining large language models (and now visual generation models) and compiling them into highly optimized inference engines that run on NVIDIA GPUs, with specialized kernels and Python/C++ runtimes.
- 68
vert.x
14.7Keclipse-vertx/vert.x · JavaVert.x Core is a toolkit for building reactive, event-driven applications on the JVM, providing low-level support for HTTP, TCP, file system access, and more.
- 69
ants
14.5Kpanjf2000/ants · Goants is a goroutine pool library for Go: it manages and recycles a massive number of goroutines with a fixed capacity, letting you limit how many run concurrently in your program.
- 70
drogon
14.3Kdrogonframework/drogon · C++Drogon is an open-source, MIT-licensed C++14/17/20 HTTP application framework for building web application server programs in C++ — from REST APIs to server-rendered sites and WebSocket apps.
- 71
memcached
14.3Kmemcached/memcached · CMemcached is a high-performance, multithreaded, event-based key/value cache store designed for use in distributed systems.
- 72
arangodb
14.3Karangodb/arangodb · C++ArangoDB is a native multi-model database that stores documents, graphs, and key-value data together, queryable with a single SQL-like language called AQL.
- 73
gel
14.2Kgeldata/gel · PythonGel is a database that builds on Postgres, adding a modern object-based data model, a graph query language called EdgeQL, built-in Auth and AI tooling, and a managed cloud service.
- 74
hugo-PaperMod
14Kadityatelange/hugo-PaperMod · HTMLHugo PaperMod is a fast, clean, responsive theme for the Hugo static site generator. It is based on hugo-paper and adds many extra features and customization options.
- 75
dask
13.9Kdask/dask · PythonDask is a flexible parallel computing library for analytics — it runs Python data workloads as scheduled tasks so they can use many cores or machines instead of a single process.
- 76
go
13.9Kjson-iterator/go · GoA high-performance JSON library for Go, positioned as a 100% compatible drop-in replacement for the standard encoding/json package: you swap the import and keep your code.
- 77
sidekiq
13.6Ksidekiq/sidekiq · RubySidekiq is a background job system for Ruby: it runs slow work like emails or uploads out-of-band, using threads to handle many jobs at the same time in the same process.
- 78
tengine
13.4Kalibaba/tengine · CTengine is a high-performance web server and reverse proxy originated by Taobao, built as a fork of nginx that stays 100% compatible with it while adding extra enterprise features.
- 79
rayon
13.4Krayon-rs/rayon · RustRayon is a data-parallelism library for Rust. It is extremely lightweight and makes it easy to turn a sequential computation into a parallel one, with data-race freedom guaranteed.
- 80
ninja
13.3Kninja-build/ninja · C++Ninja is a small build system with a single focus: speed. It is a program that runs your project's build steps — compiling and linking — quickly.
- 81
akka-core
13.3Kakka/akka-core · ScalaAkka is a platform for building highly responsive, resilient and scalable distributed services; this repo holds its core libraries. It is used for transactional, analytical, agentic-AI, digital-twin, IoT and edge-to-cloud systems.
- 82
capnproto
13.2Kcapnproto/capnproto · C++Cap'n Proto is an insanely fast data interchange format and capability-based RPC system. Think JSON, except binary — or think Protocol Buffers, except faster.
- 83
FlashMLA
13.1Kdeepseek-ai/FlashMLA · C++FlashMLA is DeepSeek's library of highly optimized CUDA attention kernels (Multi-head Latent Attention) that power the DeepSeek-V3 and DeepSeek-V3.2 models, covering both dense and sparse attention for prefill and decoding.
- 84
libfacedetection
12.8KShiqiYu/libfacedetection · C++libfacedetection is an open source C++ library for CNN-based face detection in images, designed to run extremely fast, reaching up to 1000 FPS on suitable hardware.
- 85
redpanda
12.6Kredpanda-data/redpanda · C++Redpanda is a streaming data platform that is wire-compatible with the Apache Kafka API but is built from the ground up in C++ to be lighter, faster, and simpler to operate.
- 86
concurrentqueue
12.5Kcameron314/concurrentqueue · C++moodycamel::ConcurrentQueue is an industrial-strength, lock-free multi-producer multi-consumer queue for C++11.
- 87
KeyDB
12.5KSnapchat/KeyDB · C++KeyDB is a multithreaded fork of Redis focused on performance, memory efficiency, and high throughput, offering extra features like Active Replication and Subkey Expires.
- 88
cupy
12.4Kcupy/cupy · PythonCuPy is a NumPy/SciPy-compatible array library for GPU-accelerated computing with Python: you write ordinary Python array code, and it executes on NVIDIA or AMD GPUs.
- 89
twemproxy
12.3Ktwitter/twemproxy · CTwemproxy (nutcracker) is a fast, lightweight proxy that sits in front of memcached and redis servers, reducing the number of connections to the caching backend.
- 90
taskflow
12.2Ktaskflow/taskflow · C++Taskflow is a C++ library that helps you write high-performance task-parallel programs with high programming productivity, letting you build and run graphs of dependent tasks across CPU threads and GPUs.
- 91
lz4
12.1Klz4/lz4 · CLZ4 is an open-source lossless compression algorithm and its reference C implementation, built for extreme speed: 500+ MB/s compression per core, a multi-GB/s decoder, scaling across CPU cores.
- 92
LeetCUDA
12.1Kxlite-dev/LeetCUDA · CudaLeetCUDA is an open-source book of CUDA learning notes for beginners, paired with a library of 200+ hand-written CUDA kernels (elementwise ops, HGEMM, Flash-Attention, etc.) and 100+ CUDA/LLM blog posts.
- 93
uvloop
11.9KMagicStack/uvloop · Cythonuvloop is a fast, drop-in replacement for Python's built-in asyncio event loop, built on top of the libuv library.
- 94
lua-nginx-module
11.8Kopenresty/lua-nginx-module · CIt embeds the Lua scripting language directly into NGINX, letting you handle HTTP requests with Lua code using directives like 'content_by_lua_block' instead of relying only on plain config or custom C modules.
- 95
JoltPhysics
11.7Kjrouwe/JoltPhysics · C++Jolt Physics is a multi-core friendly rigid body physics and collision detection library written in C++, built for games and VR applications. It's the engine behind 'Horizon Forbidden West' and 'Death Stranding 2: On the Beach'.
- 96
h2o
11.6Kh2o/h2o · CH2O is a new-generation HTTP server written in C that supports HTTP/1.x, HTTP/2, and experimental HTTP/3, built to be faster and more responsive than older-generation web servers.
- 97
workerman
11.5Kwalkor/workerman · PHPWorkerman is an asynchronous event-driven PHP framework for building fast, scalable network applications, with built-in support for HTTP, WebSocket, TCP, and custom protocols.
- 98
Newtonsoft.Json
11.3KJamesNK/Newtonsoft.Json · C#Json.NET (Newtonsoft.Json) is a high-performance JSON framework for .NET that serializes and deserializes .NET objects to and from JSON text.
- 99
gnet
11.2Kpanjf2000/gnet · Gognet is a high-performance, lightweight, non-blocking, event-driven networking framework written in pure Go, built on epoll and kqueue to work at the transport layer with TCP, UDP, and Unix Domain Socket.
- 100
numba
11.2Knumba/numba · PythonNumba is a Just-In-Time compiler that translates a subset of Python and NumPy code into fast machine code using LLVM, letting numerical functions run near C/Fortran speed without leaving Python.
Showing the 100 most-starred of 141. See every one in the catalog.
Related topics
Often tagged together