Special Resource Requirements

1 How many FPGA-hours would you like to request?
2 Please explain why you need access to the FPGAs
3 How many GPU-hours would you like to request?
4 Which applications are you planning to run on the GPUs? Also justify the requested amount of GPU-hours.

How many FPGA-hours would you like to request?

Please explain why you need access to the FPGAs

Provide a short description, how you intend to use the FPGAs.

How many GPU-hours would you like to request?

On the GPU partition of Noctua 1 each node has 40 CPU cores and is equipped with either one NVIDIA 1080 TI or one NVIDIA 2080 TI.
( Keep in mind that the NVIDIA 1080 TI has no Tensor Cores ).
On the GPU partition of Noctua 2 each node has 128 CPU cores and is equipped with 4 NVIDIA A100 GPUs.
Node sharing is enabled on all GPU nodes.

Example:
If you run a job on 2 GPUs and 2 CPU cores for 1 hour, you will consume:

CPU hours depend on the amount of CPU memory which has been allocated. Detailed information can be found here https://uni-paderborn.atlassian.net/wiki/spaces/PC2DOK/pages/12944197/CPU+resources#How-much-computing-time-is-required-for-your-project-in-the-upcoming-compute-time-period-of-one-year?
2 GPU hours

Which applications are you planning to run on the GPUs? Also justify the requested amount of GPU-hours.

Specify what applications you are planning to run.

Justify the requested amount of gpu-hours.
Example: Your application generates new values iteratively. A single iteration requires ~6min of GPU time. For your statistical calculations you need 10.000 values.
Hence, in order to generate 10.000 values, you need 1.000 GPU-hours (6*10.000/60).