- HPC Buyers Guide
Detailed Buyers Reference- Applications
- Budget
- Processors
- Interconnects
- Distributions
- Cluster Management
- Warranty and Support
- Cluster Racking
- Cluster Logical Layout
- Master Node
- Storage
- Backup
- Nodes
- Power
- Cooling
- Commercial Software
- Schedulers
- Account Management
- Security
- Windows Integration
- Shipping and Delivery
- On-Site and Training
- Next Steps
HPC Applications- Processor
- System
- Network
- Storage
- Power
- Cooling
- Facilities Estimator
- Remote Management
- Green Computing
- Visualization
- Compilers
- Operating Systems
- Schedulers and Tools
- Virtualization
Memory
What are your memory options?
The amount of memory needed per node is also based on your application requirements. Some applications are very memory intensive and require quite large amounts of memory per processing core, which other applications can run comfortably in 512 MB memory per core. Aspen recommends that a general use HPC cluster be configured with a minimum of 1 GB memory per core, which comfortably accommodates most HPC applications. Consult your vendor, user group, or your Aspen Systems Sales engineer for specific information about the applications you wish to run on your new cluster.
To control costs, you may also configure one or several nodes with a larger amount of memory per core to be used for those applications that require it, and configure the remaining nodes with a lesser amount of memory. Using your scheduling system or other methods, you can direct codes which require larger memory footprints to those few nodes that are equipped appropriately, and route applications with smaller memory requirements to other nodes.
Front-end nodes on your cluster are nodes which users log into to compile codes (if needed), run their applications, or perform other memory intensive operations. Larger numbers of users on the cluster can expand the memory requirements on the interactive node(s). Aspen recommends a minimum of 2 GB per core on multi-user front-end and interactive nodes.
