I have access to the Tier 1 cluster Hortense.
Hortense consists of the following partitions:

dodrio/cpu_rome: phase 1 main partition:

342 workernodes, each with:

2x 64-core AMD Epyc 7H12 CPU 2.6 GHz (“Rome” microarchitecture, 128 cores per node)

256 GiB RAM (~2GB/core), no swap

480 GB SSD local disk

dodrio/cpu_rome_512: large-memory partition:

42 workernodes, each with:

2x 64-core AMD Epyc 7H12 CPU 2.6 GHz (128 cores per node)

512 GiB RAM (~4GB/core), no swap

480 GB SSD local disk

dodrio/cpu_milan: phase 2 main partition:

384 workernodes, each with:

2x 64-core AMD Epyc 7763 CPU 2.45 GHz (“Milan” microarchitecture, 128 cores per node)

256 GiB RAM (~2GB/core), no swap

480 GB SSD local disk

dodrio/gpu_rome_a100_40: GPU partition:

20 workernodes, each with:

2x 24-core AMD Epyc 7402 CPU 2.8 GHz (48 cores per node)

4x NVIDIA A100-SXM4 (40 GB GPU memory), NVLink3

256 GiB RAM (~5GB/CPU core), no swap

480 GB SSD local disk

dodrio/gpu_rome_a100_80: phase 2 GPU partition:

20 workernodes, each with:

2x 24-core AMD Epyc 7402 CPU 2.8 GHz (48 cores per node)

4x NVIDIA A100-SXM4 (80 GB GPU memory), NVLink3

512 GiB RAM (~10GB/CPU core), no swap

480 GB SSD local disk

dodrio/debug_rome: interactive and debug partition:

3 workernodes, each with:

12-core AMD Epyc 7402 CPU 2.8 GHz (48 oversubscribed cores as seen by scheduler)

1 shared NVIDIA Quadro P1000 (4 GB GPU memory)

1 NVIDIA V100 (16 GB GPU memory)

256 GiB RAM (~5.2GB/oversubscribed core), no swap

100 GB SSD local disk