performance-study

KO | EN

gitlite — search

#python #java #python3 #arduino #golang #machine-learning #rust #html #flask #javascript #seismology #nodejs

performance-study

★ 13 Open GitHub ↗

Performance study for HPC applications across Google, AWS, and Azure clouds ☁️

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Performance Study

[![DOI](https://zenodo.org/badge/837429553.svg)](https://zenodo.org/doi/10.5281/zenodo.13738495)

This study tested HPC application performance across three clouds and on-premises HPC. The repository is organized as follows:

- [docker](docker): includes container builds for different environments. Containers are shared between environments when possible to reduce redundancy.
  - [google](docker/google): includes Google builds for each of CPU an GPU
  - [aws](docker/aws): includes AWS builds for each of CPU and GPU. The distinguishing feature is building with libfabric for EFA.
  - [azure](docker/azure): includes Microsoft Azure builds for each of CPU and GPU, targeting infiniband.

- [experiments](experiments): are organized first by cloud, and then the underlying environment. In each, a README with the full experiment protocol (and usually commands to run) are included.
  - [Google Cloud](experiments/google) includes HPC Toolkit (Compute Engine), and GKE (Kubernetes) for each of CPU and GPU
  - [Amazon Web Services](experiments/aws) includes Parallel Cluster (EC2), and EKS (KUbernetes) for each of CPU and GPU
  - [Microsoft Azure](experiments/azure) includes CycleCloud (VMs), and AKS (Kubernetes) for each of CPU and GPU.
- [analysis](analysis): includes preliminary plots for data exploration. Note that not all are finalized.
- [paper](paper): includes a subset of cleaned up and further worked on plots intended for use in publications, etc.

## Experiments

### "Bare Metal"

- [x] Microsoft Azure CycleCloud CPU (date)
  - [x] size 32 (abhik done 6 apps 8/28/2024, done milroy 8/30/2024)
  - [x] size 64 (abhik done 6 apps 8/28/2024, done milroy 8/30/2024)
  - [x] size 128 (done milroy 8/30/2024)
  - [x] size 256 (done milroy 8/31/2024)
- [x] Microsoft Azure CycleCloud GPU (date)
  - [x] size 4 (milroy and ani 8/31/2024)
  - [x] size 8 (milroy and ani 8/31/2024)
  - [x] size 16 (milroy and ani 8/31/2024)
  - [x] size 32 (milroy and ani 8/31/2024)
- [ ] AWS GPU Parallel Cluster
  - [ ] size 32 (not going to do, could not build image)
  - [ ] size 64 (not going to do, could not build image)
  - [ ] size 128 (not going to do, could not build image)
  - [ ] size 256 (not going to do, could not build image)
- [x] AWS CPU Parallel Cluster
  - [x] size 32 (done milroy 8/29/2024-8/30/2024)
  - [x] size 64 (done ani 8/29/2024-8/30/2024)
  - [x] size 128 (done ani 8/29/2024-8/30/2024)
  - [x] size 256 (done ani 8/29/2024-8/30/2024)
- [x] Google Cloud Compute Engine CPU (redone several times due to app configurations)
  - [x] size 32 (vsoch done 8/26/2024)
  - [x] size 64 (vsoch done 8/26/2024)
  - [x] size 128 (vsoch done 8/27/2024)
  - [x] size 256 (vsoch done 8/27/2024)
- [x] Google Compute Engine GPU
  - done on llnl-flux
  - [x] New VM and automation needed with Terraform (vsoch, early 9/2024)
  - [x] size 4 (vsoch 9/6/2024)
  - [x] size 8 (vsoch 9/7/2024)
  - [x] size 16 (vsoch 9/8/2024)
  - [x] size 32 (vsoch 9/8/2024)
  - [x] quicksilver and osu all reduce need runs at all sizes (vsoch 9/9/2024)

### Kubernetes

- [x] Microsoft Azure AKS CPU
  - [x] size 32 (vsoch done 8/24/2024), redone with placement (vsoch 8/28/2024)
  - [x] size 64 (vsoch done 8/24/2024), redone with placement (vsoch 8/28/2024)
  - [x] size 128 (vsoch done 8/28/2024)
  - [x] size 256 (vsoch TBA 8/29/2024)
- [x] Google Cloud GKE CPU
  - [x] size 32 (vsoch done 8/21/2024)
  - [x] size 64 (vsoch done 8/22/2024)
  - [x] size 128 (vsoch done 8/23/2024)
  - [x] size 256 (vsoch done 8/23/2024)
- [x] AWS CPU EKS
  - [x] size 32 (vsoch done 8/21/2024-8/22/2024)
  - [x] size 64  (vsoch done 8/22/2024) 
  - [x] size 128 (vsoch done 8/22/2024) 
  - [x] size 256 (vsoch done on 8/31/2024)
- [x] AWS GPU EKS
  - [x] size 4 (done vsoch 8/26/2024, milroy lammps/osu 8/27/2024)
  - [x] size 8 (done vsoch 8/26/2024, milroy lammps/osu 8/27/2024)
  - [x] size 16 (done vsoch, milroy lammps/osu 8/27/2024)
  - size 32 not possible, could not get more than 16 nodes from AWS
- [x] Google Cloud GKE GPU
  - [x] size 4 (done vsoch 8/29/2024)
  - [x] size 8 (done vsoch TBA 8/29/2024)
  - [x] size 16 (done vsoch 8/30/2024)
  - [x] size 32 (done vsoch 8/30/2024)
   - milroy figured out installing latest drivers - key to success here!
- [x] Microsoft Azure AKS GPU
  - [x] size 4 (done vsoch 8/31/2024)
  - [x] size 8 (done vsoch 8/31/2024)
  - [x] size 16 (done vsoch 8/31/2024)
  - [x] size 32 (done vsoch 8/31/2024)

## License

HPCIC DevTools is distributed under the terms of the MIT license.
All new contributions must be made under this license.

See [LICENSE](https://github.com/converged-computing/cloud-select/blob/main/LICENSE),
[COPYRIGHT](https://github.com/converged-computing/cloud-select/blob/main/COPYRIGHT), and
[NOTICE](https://github.com/converged-computing/cloud-select/blob/main/NOTICE) for details.

SPDX-License-Identifier: (MIT)

LLNL-CODE- 842614

performance-study

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README