repowiserepowise

App

aliyun / simai

★ 1,301stars on GitHubPythonApache-2.0

This connects Repowise to aliyun/simai's code. It does not install simai.

Repowise maps public repos from their source. Not affiliated with aliyun. What is Repowise?

github.com/aliyun/simaiIndexed at cef83f9

What is simai?

SimAI is a simulator for large-scale AI inference and training, used to estimate how long AI tasks take and to analyze process details. It models the framework, collective communication and network layers, so users can measure end-to-end performance under different settings such as collective algorithms, NCCL environment variables or congestion control. The core is written in Python and C++. To install it, run git clone https://github.com/aliyun/simai.git and follow the steps below. Its code health is 6.8 out of 10 (Fair), ranked #146 of 4,000+ indexed repos. The last commit to simai was on Jul 10, 2026, and 20 people have contributed.

As of commit cef83f9, indexed Sep 15, 2026. What it is comes from a model reading the README; the rest is measured from the source and git history, not from the maintainer.

For agentssimai.md

Repo facts: type, license, health, rank, contributors
simai facts
TypeApp
LicenseApache-2.0
Code health6.8 / 10 (Fair)
Health rank#146 of 4,000+ indexed repos
Last commitJul 10, 2026
Contributors20
Bus factor1
Size340 files, 32,175 lines

Install simai

Commands copied from the README

  1. 1Clone the simai repository.
    git clone https://github.com/aliyun/simai.git
  2. 2Move into the simai folder.
    cd simai
  3. 3Initialize the submodules in the repository.
    git submodule update --init --recursive
  4. 4Compile SimAI-Analytical.
    ./scripts/build.sh -c analytical
  5. 5Compile SimAI-Simulation with ns3 support.
    ./scripts/build.sh -c ns3
  6. 6Run SimAI-Analytical on the example workload, which calculates bus bandwidth automatically.
    ./bin/SimAI_analytical -w ./example/workload_analytical.txt -g 9216 -nv 360 -nic 48.5 -n_p_s 8 -g_p_s 8 -r example-

Full install notes on GitHub

Explore simai

Ask about this code in your agent

This connects Repowise to simai's code. It does not install simai. To install simai, use the install steps.

Useful if you want to change or fork simai: your agent can ask how it is put together and get answers that cite its files.

Ask about simai

Answers come from simai's README and source, with the files they cite. Two free answers a day without an account.

Want this for your own codebase? Map your repo free

About this page

Get a page like this for your own codebase

Repowise built this map of aliyun/simai from its source and git history. It parses every symbol, draws the dependency graph, scores each file's health from complexity, duplication, coverage and churn, finds hotspots and owners in the history, and writes documentation for every module. Your coding agent reads the same index over MCP, so its answers cite real files.

Free for 2 public repos, with a page like this one. No card. See how Repowise works

Prefer local? pip install repowise && repowise init indexes your own checkout, no account needed. Star Repowise on GitHub

aliyun/simai FAQ

What is SimAI?

SimAI is a simulator for large-scale AI inference and training. It models the framework, collective communication and network layers to produce end-to-end performance data for LLM workloads.

How does SimAI work?

SimAI runs a chosen mode, such as analytical or simulation, over a workload. Its network front end and system layer execute the collective and network models, and analysis components turn the outputs into performance measurements and dashboards.

How do I install SimAI?

SimAI is built from source by initializing its submodules and running the build script with the analytical or ns3 option. The example binary SimAI_analytical can then run on the included workload file. See the install steps

What language is SimAI written in?

SimAI is written mainly in Python and C++, with some shell scripts, a Dockerfile and a Makefile.

How is the SimAI code organized?

SimAI is organized into the vidur-alibabacloud and astra-sim-alibabacloud modules. The first covers scheduling scenarios and configuration in Python, and the second covers the network front end, system and workload code in C++.

Who is SimAI for?

SimAI is for researchers who want to analyze AI inference and training process details. It helps them evaluate task time under specific conditions and measure end-to-end gains from algorithm and network changes.

What license does SimAI use?

SimAI is licensed under Apache-2.0.

Code health

6.8out of 10Fair

On the leaderboards:#40 of 67 Python repos

Key features of simai

  • Estimates the time consumption of AI tasks under specific conditions for LLM training and inference.
  • SimAI-Analytical gives fast estimates by using bus bandwidth to approximate collective communication time.
  • SimAI-Simulation provides full-stack simulation with fine-grained network communication modeling through NS3.
  • Supports end-to-end inference simulation with Prefill/Decode separation and GPU memory modeling for KV cache.
  • Evaluates end-to-end performance gains from algorithm changes such as collective communication algorithms, congestion control and adaptive routing.

How aliyun/simai works

A run starts from a chosen mode (analytical, simulation or flow-only), which selects a scenario and feeds the simulation front end. The entry point astra-sim-alibabacloud/astra-sim/network_frontend/ns3/entry.h connects the network front end to the NS3 simulator, and astra-sim-alibabacloud/astra-sim/network_frontend/analytical pairs an analytical network API with a simulation driver called AnaSim. The astra-sim-alibabacloud/astra-sim/system directory executes the collective and network models, and astra-sim-alibabacloud/astra-sim/workload defines runnable simulation layers through the Layer class. On the Python side, vidur-alibabacloud/vidur/main.py is an entry point that runs scheduling scenarios, and vidur-alibabacloud/vidur/config holds the configuration dataclasses for the models it simulates. Profiling and analysis components, including vidur-alibabacloud/vidur/config_optimizer/analyzer/dashboard/main.py, turn run outputs into performance measurements and dashboards.

The codebase has 340 files and 2,486 symbols in 4 modules, led by Python, C++ and Shell. Code health is 6.8 out of 10, rated fair.