MVAPICH :: Neuron Best Practice

MVAPICH

Home
Overview
- Overview
- Features
Download
Performance
- Job Startup
  - Performance
- MV2
  - Pt-to-Pt
  - Collectives
- MV2-Azure
  - Pt-to-Pt
- MV2-X
  - XPMEM Pt-to-Pt
  - OpenSHMEM
  - UPC
  - UPC++
- MV2-X-AWS
  - Pt-to-Pt
  - Applications
  - Collectives
- MV2-X-Azure
  - Pt-to-Pt
- MV2-GDR
  - CUDA
    - Pt-to-Pt
    - Collectives
  - ROCM
    - Pt-to-Pt
- MV2-J
  - Pt-to-Pt
  - Collectives
- MV2-MIC
  - Microbenchmarks
- MV2-Virt
  - Virtual Machine Pt-to-Pt
  - Virtual Machine Applications
  - Docker Pt-to-Pt
  - Docker Applications
  - Singularity Pt-to-Pt
  - Singularity Applications
- MV2-EA
  - Applications
Benchmarks
Tools
- OEMT
- OSU INAM
Publications
Talks
Users
Best Practices
Support
- Userguide
- Mailing Lists
- FAQ
- Help
MUG
Jobs

mvapich2

MVAPICH: MPI over InfiniBand, Omni-Path, Ethernet/iWARP, RoCE, and Slingshot

Network-Based Computing Laboratory

Below is a compilation of best practices for the MVAPICH2 library in conjunction with Neuron.

Impact of transport protocol-based tuning
NEURON on Stampede2 KNL
NEURON on Stampede2 Skylake

Contact: Dhabaleswar K. (DK) Panda
Dept of Computer Science and Engineering
2001-2024 NBCL. All rights reserved.

774 Dreese Laboratories
2015 Neil Avenue
Columbus, OH 43210