Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 3 additions & 7 deletions hardware/index.rst
Original file line number Diff line number Diff line change
Expand Up @@ -45,15 +45,11 @@ There are:

- 1x ``ghlogin`` node, containing

- 1x `NVIDIA Grace Hopper Superchip <https://www.nvidia.com/en-gb/data-center/grace-hopper-superchip/>`_ (GH200 480GB)

- 1x NVIDIA Grace aarch64 CPU @ 3.483 GHz (72 Arm Neoverse V2 cores)
- 1x NVIDIA H100 96GB with 900 GB/s NVLink-C2C
- 1x `NVIDIA Grace Superchip <https://www.nvidia.com/en-gb/data-center/grace-cpu-superchip/>`_

- 480 GB LPDDR5X RAM
- 1x Mellanox CONNECTX-7 NDR200 (100Gb/s due to existing network) InfiniBand port
- 2x NVIDIA Grace aarch64 CPU @ 3.483 GHz (72 Arm Neoverse V2 cores)

- 5x ``gh`` nodes, each containing
- 7x ``gh`` nodes, each containing

- 1x `NVIDIA Grace Hopper Superchip <https://www.nvidia.com/en-gb/data-center/grace-hopper-superchip/>`_ (GH200 480GB)

Expand Down
31 changes: 22 additions & 9 deletions usage/index.rst
Original file line number Diff line number Diff line change
Expand Up @@ -217,7 +217,7 @@ As described on the :ref:`hardware` page, Bede contains a mix of nodes using 2 C
Bede's original nodes contain Power 9 CPUs (``ppc64le``), with Nvidia Volta and Turing architecture GPUs (``SM_70`` & ``sm_75``).
Jobs in the ``gpu``, ``test`` and ``infer`` partitions will run on ``ppc64le`` architecture nodes.

The newer Grace Hopper open pilot include `NVIDIA Grace Hopper Superchips <https://www.nvidia.com/en-gb/data-center/grace-hopper-superchip/>`_ which are composed of an ARM CPU (``aarch64``) and an NVIDIA Hopper GPU (``sm_90``).
The newer Grace Hopper open pilot include compute nodes with `NVIDIA Grace Hopper Superchips <https://www.nvidia.com/en-gb/data-center/grace-hopper-superchip/>`_ which are composed of an ARM CPU (``aarch64``) and an NVIDIA Hopper GPU (``sm_90``). These have a node providing interactive login sessions allowing you to prepare your work, which is based on the `NVIDIA Grace Superchip <https://www.nvidia.com/en-gb/data-center/grace-cpu/>`_ composed of two ARM CPUs.
Jobs in the ``ghlogin``, ``gh`` and ``ghtest`` partitions will run on the ``aarch64`` architecture nodes.


Expand All @@ -226,10 +226,10 @@ Jobs in the ``ghlogin``, ``gh`` and ``ghtest`` partitions will run on the ``aarc
Connecting to the ``ghlogin`` node
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

To get an interactive login-session on a Grace-Hopper node in the ``ghlogin`` partition, you must connect to Bede's regular login nodes as usual via ssh / x2go.
To get an interactive login-session for the Grace Hopper environment in the ``ghlogin`` partition, you must connect to Bede's regular login nodes as usual via ssh / x2go.

Once connected, the ``ghlogin`` command can be used to request an interactive session on the ``ghlogin`` node.
The login environment includes shared (unrestricted) access to the Hopper GPU, and by default will provide 4 CPU cores and 16GB of RAM for 8 hours.
The login environment provides a CPU and OS environment on a Grace node, which is compatible with the Grace Hoppers but without access to a GPU. By default it will provide 4 CPU cores and 16GB of RAM for 8 hours.
Use additional srun style flags to request a different duration or resources.
You must provide your project account.

Expand All @@ -242,9 +242,20 @@ You must provide your project account.
ghlogin -A <project> --time 4:00:00 -c 8 --mem 24G


This will provide shell access to the login environment, which is a single Grace Hopper superchip.
This will provide shell access to the login environment, which is a single Grace superchip.
Access is mediated by slurm and you'll have a default of 4 cores and 1GB RAM for 8 hours (amend by adding srun style flags to the ``ghlogin`` command).
Access to the GPU in the login environment is currently unrestricted.

Despite there being no GPU in the login environment, it can be used to compile CUDA applications by loading the CUDA "stub" libraries:

.. code-block:: bash

# load cuda libraries
module load cuda/12.6.2

# make stub driver libraries available
export LD_LIBRARY_PATH=/usr/lib64:$CUDA_HOME/lib64/stubs:$LD_LIBRARY_PATH

the ``ghtest`` queue can be used to run short test programs requiring a GPU (see the Running Jobs section)


Running Jobs
Expand Down Expand Up @@ -495,15 +506,17 @@ Jobs are scheduled subject to Slurm's `Multifactor Priority Plugin <https://slur
Grace-Hopper Pilot
------------------

Bede contains 6 NVIDIA Grace-Hopper nodes.
Bede contains 1 NVIDIA Grace-CPU node and 7 NVIDIA Grace-Hopper nodes.

Each Grace-CPU node node contains a single `Grace Superchip <https://www.nvidia.com/en-gb/data-center/grace-cpu/>`_, containing two 72-core 64-bit ARM CPUs. Further details are listed on the :ref:`hardware` page.

Each Grace-Hopper node contains a single `Grace Hopper Superchip <https://www.nvidia.com/en-gb/data-center/grace-hopper-superchip/>`_, containing one 72-core 64-bit ARM CPU and one 96GB Hopper GPU with NVLink-C2C providing 900GB/s of bidirectional bandwidth between the CPU and GPU. Further details are listed on the :ref:`hardware` page.


Connecting to the ``ghlogin`` node
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

To get an interactive login-session on a Grace-Hopper node in the ``ghlogin`` partition, you must connect to Bede's regular login nodes as usual via ssh / x2go.
To get an interactive login-session for the Grace-Hopper environment in the ``ghlogin`` partition, you must connect to Bede's regular login nodes as usual via ssh / x2go.

Once connected, the ``ghlogin`` command can be used to request an interactive session on the ``ghlogin`` node.

Expand All @@ -522,10 +535,10 @@ Further details and example batch job submission scripts are provided in the :re
Software availability
~~~~~~~~~~~~~~~~~~~~~

The Grace-Hopper nodes contain ``aarch64`` architecture CPUs, rather than the ``ppc64le`` architecture CPUs in the other Bede nodes.
The nodes in the Grace-Hopper environment contain ``aarch64`` architecture CPUs, rather than the ``ppc64le`` architecture CPUs in the other Bede nodes.
As such, software built for the ``ppc64le`` CPUs will not run on the ``aarch64`` CPUs and vice versa.

The Grace-Hopper nodes are also running a newer operating (Rocky 9) system than the ppc64le nodes (RHEL 8), which can impact software availability.
The Grace-Hopper environment also features a newer operating (Rocky 9) system than the ppc64le nodes (RHEL 8), which can impact software availability.

Use ``module avail`` from the ``ghlogin`` node to list centrally provided software modules for the grace-hopper nodes.

Expand Down