Lambda Vector System (GPU system)
Overview of Lambda Vector System
- The Lambda Vectors system, provided by the Biocomputational Engineering Program at the Universities at Shady Grove, is housed in the OIT staging room of the Biomedical Sciences and Engineering (BSE) building.
- The system is available to students and faculty for research and teaching, providing high-performance computing resources for data-intensive and computationally demanding tasks.
- The system offers high-performance computational resources, featuring five NVIDIA GeForce RTX 3090 GPUs.
User Guidelines
Please review and follow the instructions in the user documentation provided to you for proper system usage.
- Use proper resource requests (number of GPUs, memory, wall time) in your job scripts.
- Log out from the system when finished. Always end your session properly to free up system resources.
- Do not rely on your account for long-term storage. Back up your files regularly to external or institutional storage.
- Student accounts remain active for one year after graduation. Faculty and staff accounts are reviewed periodically.
- Keep your login credentials private and never share them with others.
Connecting to the System
There are two ways to connect to the USG network:
- When on campus, you can connect directly through the local network using your USG credentials. You may use either the “USG” or “USG-Visitor” Wi-Fi networks.
- If you are off campus, you can connect remotely to Lambda Vector using one of the following methods:
- VMware Horizon Client
https://sg-vmsec-srv.sgrove.usmd.edu/
Once connected to the USG network, you can access the GPU systems using tools such as PuTTY, WinSCP, MobaXterm, or similar applications.
- Further details will be available in the documentation you receive after your account is created.
Software & Tools
- List available compilers (GCC, CUDA toolkit)
- Pre-installed libraries (TensorFlow, PyTorch, MPI)
- MATLAB R2024a
To request/install new software, please contact biocomp-gpu@umd.edu