Skip to content

Pilusimida/spcl_HTSIM

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ECE6383 High Speed Network Assignment2

This project is forked from SPCL version htsim which is based on UEC htsim. This project is only for course assignment at NYU Tandon ECE6383

Project Structure

├── htsim
│   ├── sim
│   │   ├── BUILD.sh
│   │   ├── datacenter
│   │   │   ├── assignment2
│   │   │   │   ├── task1
│   │   │   │   ├── …
│   │   │   ├── topologies
│   │   │   │   ├── topo_assignment2
│   │   │   ├── connection_matrices
│   │   │   │   ├── cm_assignment2
│   │   │   ├── htsim_uec  # Executable Binary File
│   │   ├── build
│   │   │   ├── parse_output
│   │   │   ├── datacenter
│   │   │   │   ├── htsim_uec  # Executable Binary File

You need to run command below to compile this project:

$ cd uet-htsim/htsim/sim/
$ chmod +x ./BUILD.sh
$ ./BUILD.sh

Project Description

This project is mainly focused on simulation of different load balancing algorithms and link failure(reduction). You can jump to folder assignment2 to see more details about the instructions


Content below is original README.md from SPCL htsim

SPCL htsim

This repository is a SPCL fork of the official UEC repository. We plan to support these features on top of the UE code:

  • ATLAHS integration to run GOAL files.
  • Implementation of REPS and SMaRTT.
  • Implementation of PCM logic (in progress).
  • Loss Recovery algorithms such as PFLD.
  • General htsim performance improvements.
  • Ability to run datacenter traces.

ATLAHS Integration

This repository also supports ATLAHS integration to run GOAL files in htsim.

A number of already collected traces can be downloaded from here.

If you want more information on ATLAHS and details about how to collect traces, you can go to the official repository.

To do so, users can specify the GOAL file simply using the -goal option instead of the typical -cm option used when running connection matrices.

For example, one run can be done using:

./htsim_uec -goal atlahs_input/llama_N4_GPU16.bin -sender_cc_only -nodes 1024 -end 10000000 -topo topologies/fat_tree_1024_1os.topo -linkspeed 200000 > output.tmp

Ongoing Development

We note that this repository is an ongoing development and more features and documentation will be added in the near future. Please feel free to report issues or problems using GitHub issues.

Purpose and Scope

HTSIM is a high-performance discrete event simulator used for network simulation. It offers faster simulation methods compared to other options, making it ideal for modeling and developing congestion algorithms and new network protocols. The role of htsim in the Ultra Ethernet Consortium (UEC) standards development is to support the transport layer working group's work on congestion control mechanisms.

In UEC, htsim:

  • provides a platform for continuous implementation and development of UEC transport layer.
  • is used to simulate and run different topologies and scenarios, helping to identify issues in the current specifications and estimate the throughput and latency for given parameters like topology, flow matrix and congestion configuration.
  • provides a reference for users and developers to run simulations with different configurable parameters for various scenarios and algorithms

htsim's role is deliberately focused on congestion control.

UEC's htsim is not:

  • a complete implementation of the UEC transport specification.
  • a standard in any way; specifically, it is not part of the official UEC standards release. While we aim to match the spec as closely as possible, there might be discrepancies between the UEC CMS specification and the simulator. Only the official CMS specification is significant, the simulator is not.

More insutrctions about htsim

Check the README file in the htsim/ folder.

References

If you use ATLAHS for your research, please cite our paper using:

@misc{shen2025atlahsapplicationcentricnetworksimulator,
      title={ATLAHS: An Application-centric Network Simulator Toolchain for AI, HPC, and Distributed Storage}, 
      author={Siyuan Shen and Tommaso Bonato and Zhiyi Hu and Pasquale Jordan and Tiancheng Chen and Torsten Hoefler},
      year={2025},
      eprint={2505.08936},
      archivePrefix={arXiv},
      primaryClass={cs.DC},
      url={https://arxiv.org/abs/2505.08936}, 
}

About

No description, website, or topics provided.

Resources

License

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 91.0%
  • Python 4.3%
  • C 3.2%
  • Shell 1.2%
  • CMake 0.3%
  • Makefile 0.0%