| Date | Event | Location | Content | Type |
|---|---|---|---|---|
| Dec 5, 2024 | Bench'24 Conference | 📍 Guangzhou | SimAI Tutorial & Deep-dive Session | 🎓 On-site |
| Nov 22, 2024 (tentative) | SimAI Community Live Stream | 🌐 Online | Interactive Technical Discussion & Demo | 💻 Virtual |
| Date | Event | Location | Content | Type |
|---|---|---|---|---|
| Nov 15, 2024 | Technical Workshop | 📍 Thousand Island Lake | SimAI Offline Technical Exchange | 🎯 On-site |
| Oct 18, 2024 | Guest Lecture | 📍 Fudan University | SimAI Tutorial & Public Course | 🎓 On-site |
| Sept 24-26, 2024 | CCF HPC China 2024 | 📍 Wuhan | SimAI Introduction & Technical Presentation | 🎤 Conference |
Don't miss our comprehensive session at Bench'24 Register Now →
|--- AICB SimAI --|--- SimCCL |--- astra-sim-alibabacloud |--- ns-3-alibabacloud
SimAI is the industry's first full-stack, high-precision Simulator for AI large-scale training. It provides detailed modeling and simulation of the entire LLM training process, encompassing framework, collective communication, network layers, and more. This comprehensive approach offers end-to-end performance data, enabling researchers to:
- Analyze training process details
- Evaluate the time consumption of AI tasks under specific conditions
- Evaluate E2E performance gains from various algorithmic optimizations including:
- Framework parameters settings
- Collective communication algorithms
- NCCL environment variables
- Network transmission protocols
- Congestion control algorithms
- Adaptive routing algorithms
- Scale-up/out network topology modifications
- ...
Building on pure simulation capabilities, SimAI has evolved into a versatile full-stack toolkit comprising four components (aicb, SimCCL, astra-sim-alibabacloud, ns-3-alibabacloud). These components can be combined in various ways to achieve different functionalities. Below, we present the six main usage scenarios for SimAI. We encourage users to explore even more possibilities with this powerful tool.
| Scenario | Description | Component Combination |
|---|---|---|
| 1. AICB Test Suite | Run communication patterns on GPU clusters using AICB Test suite | AICB |
| 2. AICB/AIOB Workload | Model compute/communication patterns of training process to generate workload | AICB |
| 3. Collective Comm Analyze | Break down collective communication operations into point-to-point communication sets | SimCCL |
| 4. Collective Comm w/o GPU | Perform RDMA collective communication traffic on non-GPU clusters | AICB + SimCCL + astra-sim-alibabacloud(physical) |
| 5. SimAI-Analytical | Conduct rapid AICB workload analysis and simulation on any server (ignoring underlying network details) | AICB + astra-sim-alibabacloud(analytical) |
| 6. SimAI-Simulation | Perform full simulation on any server | AICB + SimCCL + astra-sim-alibabacloud(simulation) + ns-3-alibabacloud |
Below is the architecture diagram of the SimAI Simulator:
astra-sim-alibabacloud is extended from astra-sim, we have integrated NCCL algorithms and added some new features.
SimAI work has been accepted by NSDI'25 Spring, for more details, please refer to our paper below:
SimAI: Unifying Architecture Design and Performance Tunning for Large-Scale Large Language Model Training with Scalability and Precision.
[pdf] / [slides] / [video]
Here are some simple examples, SimAI full tutorials can be found here: [SimAI@Tutorial], aicb@Tutorial, [SimCCL@Tutorial], [ns-3-alibabacloud@Tutorial]
You can follow the instrucitons below to quickly set up the environtments and run SimAI
The following code has been successfully tested on GCC/G++ 9.4.0, python 3.8.10 in Ubuntu 20.04
# Clone the repository
$ git clone https://github.com/aliyun/SimAI.git
$ cd ./SimAI/
# Clone submodules
$ git submodule update --init --recursive
# Make sure use the newest commit
$ git submodule update --remote
# Compile SimAI-Analytical
$ ./scripts/build.sh -c analytical
# Compile SimAI-Simulation (ns3)
$ ./scripts/build.sh -c ns3
$ ./bin/SimAI_analytical -w example/workload_analytical.txt -g 1024 -g_p_s 8 -r results/test- -busbw example/busbw.yaml# Create network topo
$ python3 ./astra-sim-alibabacloud/inputs/topo/gen_HPN_7.0_topo_mulgpus_one_link.py -g 128 -gt A100 -bw 400Gbps -nvbw 2400Gbps
# Running
$ AS_SEND_LAT=3 AS_NVLS_ENABLE=1 ./bin/SimAI_simulator -t 16 -w ./example/microAllReduce.txt -n ./HPN_7_0_128_gpus_8_in_one_server_with_400Gbps_A100
Please email Gang Lu ([email protected]) or Qingxu Li ([email protected]) if you have any questions.
Welcome to join the SimAI community chat groups, with the DingTalk group on the left and the WeChat group on the right.
