GitHub C++ Trending

The latest build: 2024-06-20Source of data: GitHubTrendingRSS

Productive, portable, and performant GPU programming in Python.

Latest ReleasedownloadsCINightly Releasediscord invitation link

pip install taichi # Install Taichi Langti gallery # Launch demo gallery

What is Taichi Lang?

Taichi Lang is an open-source, imperative, parallel programming language for high-performance numerical computation. It is embedded in Python and uses just-in-time (JIT) compiler frameworks, for example LLVM, to offload the compute-intensive Python code to the native GPU or CPU instructions.

The language has broad applications spanning real-time physical simulation, numerical computation, augmented reality, artificial intelligence, vision and robotics, visual effects in films and games, general-purpose computing, and much more.


Why Taichi Lang?

  • Built around Python: Taichi Lang shares almost the same syntax with Python, allowing you to write algorithms with minimal language barrier. It is also well integrated into the Python ecosystem, including NumPy and PyTorch.
  • Flexibility: Taichi Lang provides a set of generic data containers known as SNode (/snod/), an effective mechanism for composing hierarchical, multi-dimensional fields. This can cover many use patterns in numerical simulation (e.g. spatially sparse computing).
  • Performance: With the @ti.kernel decorator, Taichi Lang's JIT compiler automatically compiles your Python functions into efficient GPU or CPU machine code for parallel execution.
  • Portability: Write your code once and run it everywhere. Currently, Taichi Lang supports most mainstream GPU APIs, such as CUDA and Vulkan.
  • ... and many more features! A cross-platform, Vulkan-based 3D visualizer, differentiable programming, quantized computation (experimental), etc.

Getting Started


  • Operating systems
    • Windows
    • Linux
    • macOS
  • Python: 3.6 ~ 3.10 (64-bit only)
  • Compute backends
    • x64/ARM CPUs
    • CUDA
    • Vulkan
    • OpenGL (4.3+)
    • Apple Metal
    • WebAssembly (experiemental)

Use Python's package installer pip to install Taichi Lang:

pip install --upgrade taichi

We also provide a nightly package. Note that nightly packages may crash because they are not fully tested. We cannot guarantee their validity, and you are at your own risk trying out our latest, untested features. The nightly packages can be installed from our self-hosted PyPI (Using self-hosted PyPI allows us to provide more frequent releases over a longer period of time)

pip install -i https://pypi.taichi.graphics/simple/ taichi-nightly

Run your "Hello, world!"

Here is how you can program a 2D fractal in Taichi:

# python/taichi/examples/simulation/fractal.pyimport taichi as titi.init(arch=ti.gpu)n = 320pixels = ti.field(dtype=float, shape=(n * 2, n))@ti.funcdef complex_sqr(z): return ti.Vector([z[0]**2 - z[1]**2, z[1] * z[0] * 2])@ti.kerneldef paint(t: float): for i, j in pixels: # Parallelized over all pixels c = ti.Vector([-0.8, ti.cos(t) * 0.2]) z = ti.Vector([i / n - 1, j / n - 0.5]) * 2 iterations = 0 while z.norm() < 20 and iterations < 50: z = complex_sqr(z) + c iterations += 1 pixels[i, j] = 1 - iterations * 0.02gui = ti.GUI("Julia Set", res=(n * 2, n))for i in range(1000000): paint(i * 0.03) gui.set_image(pixels) gui.show()

If Taichi Lang is properly installed, you should get the animation below :

See Get started for more information.

Build from source

If you wish to try our our experimental features or build Taichi Lang for your own environments, see Developer installation.


Community activity Time period

Timeline graphIssue status graphPull request status graphTrending topics


Kudos to all of our amazing contributors! Taichi Lang thrives through open-source. In that spirit, we welcome all kinds of contributions from the community. If you would like to participate, check out the Contribution Guidelines first.

Contributor avatars are randomly shuffled.


Taichi Lang is distributed under the terms of Apache License (Version 2.0).

See Apache License for details.


For more information about the events or community, please refer to this page

Join our discussions

Report an issue

Contact us



AOT deployment

Lectures & talks


If you use Taichi Lang in your research, please cite the corresponding papers:

Atmosphère is a work-in-progress customized firmware for the Nintendo Switch.


LicenseChat on Discord

Atmosphère is a work-in-progress customized firmware for the Nintendo Switch.


Atmosphère consists of multiple components, each of which replaces/modifies a different component of the system:

  • Fusée: First-stage Loader, responsible for loading and validating stage 2 (custom TrustZone) plus package2 (Kernel/FIRM sysmodules), and patching them as needed. This replaces all functionality normally in Package1loader/NX Bootloader.
    • Sept: Payload used to enable support for runtime key derivation on 7.0.0.
  • Exosphère: Customized TrustZone, to run a customized Secure Monitor
  • Thermosphère: EL2 EmuNAND support, i.e. backing up and using virtualized/redirected NAND images
  • Stratosphère: Custom Sysmodule(s), both Rosalina style to extend the kernel/provide new features, and of the loader reimplementation style to hook important system actions
  • Troposphère: Application-level Horizon OS patches, used to implement desirable CFW features


This software is licensed under the terms of the GPLv2, with exemptions for specific projects noted below.

You can find a copy of the license in the LICENSE file.


  • Nintendo is exempt from GPLv2 licensing and may (at its option) instead license any source code authored for the Atmosphère project under the Zero-Clause BSD license.


Atmosphère is currently being developed and maintained by SciresM, TuxSH, hexkyz, and fincs.
In no particular order, we credit the following for their invaluable contributions:

  • switchbrew for the libnx project and the extensive documentation, research and tool development pertaining to the Nintendo Switch.
  • devkitPro for the devkitA64 toolchain and libnx support.
  • ReSwitched Team for additional documentation, research and tool development pertaining to the Nintendo Switch.
  • ChaN for the FatFs module.
  • Marcus Geelnard for the bcl-1.2.0 library.
  • naehrwert and st4rk for the original hekate project and its hwinit code base.
  • CTCaer for the continued hekate project's fork and the minerva_tc project.
  • m4xw for development of the emuMMC project.
  • Riley for suggesting "Atmosphere" as a Horizon OS reimplementation+customization project name.
  • hedgeberg for research and hardware testing.
  • lioncash for code cleanup and general improvements.
  • jaames for designing and providing Atmosphère's graphical resources.
  • Everyone who submitted entries for Atmosphère's splash design contest.
  • All those who actively contribute to the Atmosphère repository.

SRS is a simple, high-efficiency, real-time video server supporting RTMP, WebRTC, HLS, HTTP-FLV, SRT, MPEG-DASH, and GB28181.

SRS(Simple Realtime Server)

SRS/6.0 (Hang) is a simple, high-efficiency, and real-time video server, supporting RTMP/WebRTC/HLS/HTTP-FLV/SRT/MPEG-DASH/GB28181, Linux/Windows/macOS, X86_64/ARMv7/AARCH64/M1/RISCV/LOONGARCH/MIPS, and essential features.

SRS Overview

Note: For more details on the single-node architecture for SRS, please visit the following link.

SRS is licenced under MIT, and some third-party libraries are distributed under their licenses.


Please check the Getting Started guide in English or Chinese. We highly recommend using SRS with docker:

docker run --rm -it -p 1935:1935 -p 1985:1985 -p 8080:8080 \ -p 8000:8000/udp -p 10080:10080/udp ossrs/srs:5

Tips: If you're in China, use this image registry.cn-hangzhou.aliyuncs.com/ossrs/srs:5 for faster speed.

Open http://localhost:8080/ to verify, and then stream using the following FFmpeg command:

ffmpeg -re -i ./doc/source.flv -c copy -f flv -y rtmp://localhost/live/livestream

Alternatively, stream by OBS using the following configuration:

  • Service: Custom
  • Server: rtmp://localhost/live
  • Stream Key: livestream

Play the following streams using media players:

If you'd like to use WebRTC, convert RTMP to WebRTC, or convert WebRTC to RTMP, please check out the wiki documentation in either English or Chinese.

To learn more about RTMP, HLS, HTTP-FLV, SRT, MPEG-DASH, WebRTC protocols, clustering, HTTP API, DVR, and transcoding, please check the documents in English or Chinese.


Would you like additional assistance from us? By becoming a sponsor or backer of SRS, we can provide you with the support you need:

  • Backer: $5 per month, online text chat support through Discord.
  • Sponsor: $100 per month, online text chat plus online meeting support.

Please visit OpenCollective to become a backer or sponsor, and send us a direct message on Discord. We are currently providing support to the developers listed below:

At SRS, our goal is to create a free, open-source community that helps developers all over the world build high-quality streaming and RTC platforms for their businesses.


The authors, TOC(Technical Oversight Committee), and contributors are listed here. The TOC members who made significant contributions and maintained parts of SRS are listed below:

  • Winlin: Founder of the project, focusing on ST and Issues/PR. Responsible for architecture and maintenance.
  • ZhaoWenjie: One of the earliest contributors, focusing on HDS and Windows. Has expertise in client technology.
  • ShiWei: Specializes in SRT and H.265, maintaining SRT and FLV patches for FFmpeg. An expert in codecs and FFmpeg.
  • XiaoZhihong: Concentrates on WebRTC/QUIC and SRT, with expertise in network QoS. Contributed to ARM on ST and was the original contributor for WebRTC.
  • WuPengqiang: Focused on H.265, initially contributed to the FFmpeg module in SRS for transcoding AAC with OPUS for WebRTC.
  • XiaLixin: Specializes in GB28181, with expertise in live streaming and WebRTC.
  • LiPeng: Concentrates on WebRTC and contributes to memory management and smart pointers.
  • ChenGuanghua: Focused on WebRTC/QoS and introduced the Asan toolchain to SRS.
  • ChenHaibo: Specializes in GB28181 and HTTP API, contributing to patches for FFmpeg with WHIP.
  • ZhangJunqin: Focused on H.265, Prometheus Exporter, and API module.

A huge THANK YOU goes out to:

We're really thankful to everyone in the community for helping us find bugs and improve the project. To stay in touch and keep helping our community, please check out this guide.


FOSSA Status

SRS is licenced under MIT, and some third-party libraries are distributed under their licenses.


  • 2024-06-15, Release v6.0-d5, v6.0-d5, 6.0 dev5, v6.0.129, 168454 lines.
  • 2024-02-15, Release v6.0-d4, v6.0-d4, 6.0 dev4, v6.0.113, 167695 lines.
  • 2023-11-19, Release v6.0-d3, v6.0-d3, 6.0 dev3, v6.0.101, 167560 lines.
  • 2023-09-28, Release v6.0-d2, v6.0-d2, 6.0 dev2, v6.0.85, 167509 lines.
  • 2023-08-31, Release v6.0-d1, v6.0-d1, 6.0 dev1, v6.0.72, 167135 lines.
  • 2023-07-09, Release v6.0-d0, v6.0-d0, 6.0 dev0, v6.0.59, 166739 lines.
  • 2024-06-15, Release v5.0-r3, v5.0-r3, 5.0 release3, v5.0.213, 163585 lines.
  • 2024-04-03, Release v5.0-r2, v5.0-r2, 5.0 release2, v5.0.210, 163515 lines.
  • 2024-02-15, Release v5.0-r1, v5.0-r1, 5.0 release1, v5.0.208, 163441 lines.
  • 2023-12-30, Release v5.0-r0, v5.0-r0, 5.0 release0, v5.0.205, 163363 lines.
  • 2023-11-19, Release v5.0-b7, v5.0-b7, 5.0 beta7, v5.0.200, 163305 lines.
  • 2023-10-25, Release v5.0-b6, v5.0-b6, 5.0 beta6, v5.0.195, 163303 lines.
  • 2023-09-28, Release v5.0-b5, v5.0-b5, 5.0 beta5, v5.0.185, 163254 lines.
  • 2023-08-31, Release v5.0-b4, v5.0-b4, 5.0 beta4, v5.0.176, 162919 lines.
  • 2023-08-02, Release v5.0-b3, v5.0-b3, 5.0 beta3, v5.0.170, 162704 lines.
  • 2023-07-09, Release v5.0-b2, v5.0-b2, 5.0 beta2, v5.0.166, 162520 lines.
  • 2023-06-11, Release v5.0-b1, v5.0-b1, 5.0 beta1, v5.0.157, 162494 lines.
  • 2023-05-14, Release v5.0-b0, v5.0-b0, 5.0 beta0, v5.0.155, 162600 lines.
  • 2023-03-23, Release v5.0-a5, v5.0-a5, 5.0 alpha5, v5.0.148, 162066 lines.
  • 2023-02-12, Release v5.0-a4, v5.0-a4, 5.0 alpha4, v5.0.141, 161897 lines.
  • 2023-01-02, Release v5.0-a3, v5.0-a3, 5.0 alpha3, v5.0.128, 161327 lines.
  • 2022-12-18, Release v5.0-a2, v5.0-a2, 5.0 alpha2, v5.0.112, 161233 lines.
  • 2022-12-01, Release v5.0-a1, v5.0-a1, 5.0 alpha1, v5.0.100, 160817 lines.
  • 2022-11-25, Release v5.0-a0, v5.0-a0, 5.0 alpha0, v5.0.98, 159813 lines.
  • 2022-11-22, Release v4.0-r4, v4.0-r4, 4.0 release4, v4.0.268, 145482 lines.
  • 2022-09-16, Release v4.0-r3, v4.0-r3, 4.0 release3, v4.0.265, 145328 lines.
  • 2022-08-24, Release v4.0-r2, v4.0-r2, 4.0 release2, v4.0.257, 144890 lines.
  • 2022-06-29, Release v4.0-r1, v4.0-r1, 4.0 release1, v4.0.253, 144680 lines.
  • 2022-06-11, Release v4.0-r0, v4.0-r0, 4.0 release0, v4.0.252, 144680 lines.
  • 2020-06-27, Release v3.0-r0, 3.0 release0, 3.0.141, 122674 lines.
  • 2020-02-02, Release v3.0-b0, 3.0 beta0, 3.0.112, 121709 lines.
  • 2019-10-04, Release v3.0-a0, 3.0 alpha0, 3.0.56, 107946 lines.
  • 2017-03-03, Release v2.0-r0, 2.0 release0, 2.0.234, 86373 lines.
  • 2016-08-06, Release v2.0-b0, 2.0 beta0, 2.0.210, 89704 lines.
  • 2015-08-23, Release v2.0-a0, 2.0 alpha0, 2.0.185, 89022 lines.
  • 2014-12-05, Release v1.0-r0, all bug fixed, 1.0.10, 59391 lines.
  • 2014-10-09, Release v0.9.8, all bug fixed, 1.0.0, 59316 lines.
  • 2014-04-07, Release v0.9.1, live streaming. 30000 lines.
  • 2013-10-23, Release v0.1.0, rtmp. 8287 lines.
  • 2013-10-17, Created.


Please read FEATURES.


Please read CHANGELOG.


Please read PERFORMANCE.




Please read PORTS.


Please read APIS.


Please read MIRRORS.


Please read DOCKERS.

Beijing, 2013.10