Type 1 – Interactive HPC (UCloud) 
¶
The easiest-to-use HPC service is DeiC Interactive HPC (Type 1) also known as UCloud.
This service is provided by the Danish universities SDU and AAU and consist of three HPC systems:
| Institution | System | User Access | User-Friendly | Sensitive data |
|---|---|---|---|---|
| SDU/K8 | Kubernetes | GUI / Interactive | Yes | Yes |
| AAU/K8 | Kubernetes | GUI / Interactive | Yes | Yes |
| AAU | Open Stack | SSH / terminal | No | No |
Costs¶
Machine types are devided into CPU and GPU based machines with their costs being measured in core hours and GPU core-hours respectively.
A 1 (GPU) core machine cost 1 (GPU) core-hour pr. hour and a 64 (GPU) core machine cost 1 (GPU) core-hour pr. hour.
Machine Type¶
These HPC systems provides users with the following machine types:
| Name | Institution | CPU (Cores) | RAM (GB) | GPU (Cores) | User Access | User-Friendly |
|---|---|---|---|---|---|---|
| u1-standard | SDU/K8 | 64 | 384 | N/A | GUI / Interactive | Yes |
| u3-gpu | SDU/K8 | 192 | 768 | 4x NVIDIA H100 | GUI / Interactive | Yes |
| uc1-gc | AAU/K8 | 128 | 512 | N/A | GUI / Interactive | Yes |
| uc1-A10 | AAU/K8 | 64 | 256 | 4x NVIDIA A10 | GUI / Interactive | Yes |
| uc-t4 | AAU | 60 | 240 | 6x Nvidia T4 | SSH / terminal | No |
| uc-a10 | AAU | 64 | 240 | 4x Nvidia A10 | SSH / terminal | No |
| uc-a40 | AAU | 64 | 480 | 3x Nvidia A40 | SSH / terminal | No |
| uc-a100 | AAU | 64 | 480 | 2x Nvidia A100 | SSH / terminal | No |
(SDU/K8) provides CPU and GPU based containerized applications such as MATLAB, STATA, RStudio, and JupyterLab through a graphical user interface (GUI), in the same way as they would on your laptop. See all apps.
CPU Nodes (u1-standard):
Dell PowerEdge C6420
64 Core Intel Xeon Gold 6130
384 GB DDR4-2400
GPU Nodes (u3-gpu):
Lenovo ThinkSystem SR675 V3
192 Core AMD EPYC 9454
768 GB DDR5-4800
4x NVIDIA Hopper H100-SXM5, 80 GB
Storage:
u1-cephfs - The storage system for DeiC interactive HPC
(AAU/K8) provides CPU and GPU based containerized applications such as MATLAB, STATA, RStudio, and JupyterLab through a graphical user interface (GUI), in the same way as they would on your laptop. See all apps.
CPU Nodes (uc1-gc):
128 Core AMD EPYC 7702
512 GBs of RAM
GPU Nodes (uc1-A10):
64 Core Intel Xeon Gold 6326
256 GBs of RAM
4x NVIDIA A10 GPU
Storage:
uc1-cephfs - The storage system for DeiC interactive HPC (AAU).
AAU provides primary GPU based virtual machines. Access is obtained through terminal and SSH. It is possible to set up interactive enviroments such as JupyterLab.
Virtual machines hosted by AAU are only suitable for the processing of
- public and,
- low risk internal data.
The specific data classification model applicable to your institution should be consulted before using these compute resources.
AUTOMATIC SHUT OFF AND DELETION
It is your responsibility to ensure that you have sufficient resources to keep your virtual machine running while you need it. All virtual machines are automatically shut off when the user runs out of resources.
To restart the machine, additional funds should be requested from your front office to cover the full duration that the virtual machine is shut off, plus any additional compute time.
All data and information related to virtual machines are IRRECOVERABLY deleted if shut off for:
- A100 virtual machines: 3 days (72 hours)
- All other virtual machines: 14 days (336 hours)
Four different machine types based on different Nvidia GPUs (T4, A10 , A40 and A100) with different application purposes.
Nvidia T4 is avaliable on the following machines:
Nvidia A10 is avaliable on the following machines:
Nvidia A40 is avaliable on the following machines:
Nvidia A100 is avaliable on the following machines:
