Computing Resources
DSI provides the following computing resources to members of the Institute.
To learn more about becoming a member, click here.
Servers
A high-end server is available for administrative, research, and teaching purposes. The top level specs are:
- 2x Intel Xeon E5-2620 CPUs @ 2.10GHz (8 cores/16 threads each)
- 4x Nvidia GTX 1080 Ti GPUs @ 1582MHz (3584 CUDA cores each)
- 128GB ECC RAM memory
- 400GB SSD (RAID 1) system drives
- 30TB HDD (RAID 5) mass storage – available upon request
- Ubuntu 18.04 Linux operating system
This system is intended for machine learning and similar applications, and includes the following software already installed:
- Google TensorFlow
- Nvidia CUDA drivers (v430.x), Toolkit (v10.x) and cuDNN (v6.x/7.x)
- C (v7.4x), Python (v2.x/3.x) and R (v3.x) programming languages
It also has a highly-performant web stack installed for hosting static or dynamic websites, consisting of the following components:
- Nginx (SSL termination)
- Varnish (cache)
- Apache (web server)
- PHP (application layer)
- Redis (key store)
- MySQL (database)
Other software or packages may be installed based on your needs. Email datascience-it@columbia.edu and you will be provided with a local Unix account based on your UNI.
Habañero HPC
On the Habañero High-Performance Cluster (HPC) maintained by Columbia University, the Institute has dedicated resources including:
- 2x high memory nodes (512 GB each)
- 6x GPU nodes (2x Nvidia K80/4x P100 GPUs each)
- 30TB scratch space
Habañero cluster documentation may be found HERE. Email hpc-support@columbia.edu with a copy to datascience-it@columbia.edu for approval.
Questions and Support
Requests should be forwarded to datascience-it@columbia.edu with a brief subject line containing the nature of the request.
CUIT Resources
Software available to the Columbia University community, for free or at Columbia’s negotiated rate may be found HERE.