# aliyun/SimAI

Repository: https://github.com/aliyun/SimAI
Canonical: https://ross.abutalabs.com/products/simai
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-08-14T09:18:27+00:00

## Health v2 (maintenance only)
Score: 66/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 97, release rhythm 35, longevity 49
- inputs: {"age_days": 691, "days_push": 19, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1133, forks 178 (observed 2026-08-28T04:03:42.925658+00:00)

## What it is
SimAI is a large-scale network simulation toolkit from Alibaba Cloud for modeling AI training and inference workloads on GPU clusters, published at NSDI'25. It combines workload generation (AICB), collective communication modeling (SimCCL), and request scheduling (adapted from Vidur) to simulate end-to-end distributed LLM training and prefill/decode inference.

## Use cases
- simulate network traffic of large-scale LLM training on GPU clusters
- evaluate network designs for distributed AI training before deployment
- model prefill/decode disaggregated inference performance
- estimate GPU memory and decode latency for LLM inference
- generate training workloads for models like DeepSeek and Qwen
- analyze collective communication flows offline with SimCCL

## When to choose
- you are designing or evaluating network topologies and congestion control for AI clusters
- you need to predict training or inference performance without running real GPU jobs
- you want to study collective communication behavior at scale

## When to avoid
- you need a production training or serving framework rather than a simulator
- you want quick single-GPU model benchmarking
- you need cycle-accurate hardware simulation

## Facets
- artifact type: library
- maturity: active
- function: simulation, machine-learning, networking, benchmarking, llm-training, llm-inference
- domain: simulation, machine-learning, networking, microservices, large-language-models, performance
- platform: python
- tags: network-simulation, gpu-clusters, collective-communication, nsdi-paper, training-simulation, inference-simulation, alibaba-cloud, linux

## Member repositories
- aliyun/SimAI (main) score 66

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:42.925658+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:37:24.727487+00:00, confidence not recorded.
  - readme: https://github.com/aliyun/SimAI (fetched 2026-08-28T04:03:42.925658+00:00, sha baa4d4f48a1e)
- Data as of 2026-08-30T08:39:29.467469+00:00.