RCAC - Knowledge Base: Bell User Guide: How do I know Non-uniform Memory Access (NUMA) layout on Bell?

Rosen Center for Advanced Computing

Compute
Bell User Guide
- Overview of Bell
- Biography of Clara Bell Sessions
- Accounts
- File Storage and Transfer
- Gateway (Open OnDemand)
- Software
- Compiling Source Code
- Running Jobs
- Frequently Asked Questions
  - About Bell
  - Logging In & Accounts
  - Jobs
    - Errors
    - Questions
      - How do I know Non-uniform Memory Access (NUMA) layout on Bell?
      - Why cannot I use --mem=0 when submitting jobs?
      - Can I extend the walltime on a job?
  - Data
  - Software
  - About Research Computing
- Datasets
Gilbreth User Guide
Weber User Guide
Scholar User Guide
Hammer User Guide
Negishi User Guide
Geddes User Guide
Anvil User Guide
Gautschi User Guide
Datasets
iGenomes
Software Catalog
Compilers
MPIs
Applications
Utilities
Biocontainers
NVIDIA NGC containers
AMD ROCm containers
FAQs
Storage
Data Depot User Guide
Fortress User Guide
REED Folder User Guide
Box Research Lab Folder User Guide
Scratch User Guide
Home Directory User Guide
Services
High-Performance Computing
Services Guides
Slurm
Depot Object User Guide
Rowdy User Guide
Environment Management with the Module Command
Protected Data Filesystem User Guide
Protected Data Archive User Guide
Purdue GenAI Studio
Environment Management with the Module Command
Environment Management with the Module Command

Expand Topics

How do I know Non-uniform Memory Access (NUMA) layout on Bell?

You can learn about processor layout on Bell nodes using the following command:
```
bell-a003:~$ lstopo-no-graphics
```

For detailed IO connectivity:

bell-a003:~$ lstopo-no-graphics --physical --whole-io

Please note that NUMA information is useful for advanced MPI/OpenMP/GPU optimizations. For most users, using default NUMA settings in MPI or OpenMP would give you the best performance.

Helpful?