Skip to content

Move and review Node Features section for Alpine #596

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
b-reyes wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

Move and review Node Features section for Alpine #596

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
35 changes: 35 additions & 0 deletions docs/clusters/alpine/alpine-hardware.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -239,6 +239,41 @@ $ sinfo --Format Partition,Gres |grep gpu

`````

### Node Features

The Alpine cluster features some heterogeneity with respect to compute node resources. To enable users to apply fine grained selection of particular resources, we set Slurm features on each compute node. These resource flags specify items such as CPU model, physical rack, and Infiniband capability. If you would like to see all the features that are available on every node in the cluster, you can use the `sinfo` command.

```bash
[ralphie@login-ci4 ~]$ sinfo --format="%N | %f"
```
Additionally, if you would like to see the features applied to a specific node, you can use `scontrol show node <node name>` and refer to the `ActiveFeatures` setting. For example, we can see the active features on the compute node `c3cpu-e2-u2` as follows:
```bash
[ralphie@login-ci4 ~]$ scontrol show node c3cpu-e2-u2 | grep ActiveFeatures
ActiveFeatures=hpcf,e2,ucb,cpu,amd-cpu,genoa,epyc-9534,128c,1t,ib
```

Features on Alpine follow the following format:
```
ActiveFeatures=<data center>,<rack>,<institution>,<node type>,<CPU chip manufacturer>-cpu,<CPU microarchitecture generation>,<cpu model>,<num cores>c,<num threads per core>t, <GPU manufacturer>-gpu,ib
```

In the table below we provide descriptions and examples for each of these features.

| Feature | Description | Example |
| -------- | ----------- | ------- |
| `<data center>` | The data center where the node is located. | `hpcf`, `spsc` |
| `<rack>` | The rack the node is located in. | `e2`, `c9`, `a9` |
| `<institution>` | The institution the node belongs to. | `ucb`, `csu`, `amc`, `rmacc` |
| `<node type>` | The defining hardware type for the node. | `cpu`, `mem`, `gpu` |
| `<CPU chip manufacturer>-cpu` | The company that designs the CPU chip. | `intel-cpu`, `amd-cpu` |
| `<CPU microarchitecture generation>` | The CPU microarchitecture generation for the chip i.e. code name. | `milan`, `rome`, `genoa`, `turin`, `grace` |
| `<cpu model>` | The CPU’s processor model. | `7313`, `7443`, `74F3`, `7502`, `7543`, `7713`, `7713P`, `9534`, `9555`, `grace-a02` |
| `<num cores>c` | The total number of cores on the node. | `128c`, `64c`, `48c` |
| `<num threads per core>t` | The number of threads per core. | `1t`, `2t` |
| `<GPU manufacturer>-gpu` | The company that designs the GPU. Note that this will be left off it is not a GPU node. | `nvidia-gpu`, `amd-gpu` |
| `<ib, if Infiniband is present>` | States if the node has Infiniband and is connected to other nodes via Infiniband. If there is no Infiniband on the node, this will be left off. | `ib` |


# Special-Purpose Resources

To help users test out their workflows, CURC provides several special-purpose resources on Alpine. These resources enable users to quickly test or compile code on CPU and GPU compute nodes. To ensure equal access to these resources, the amount of resources (such as CPUs, GPUs, and runtime) are limited.
Expand Down
24 changes: 0 additions & 24 deletions docs/clusters/alpine/quick-start.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,30 +48,6 @@ The Alpine cluster has different types of interconnects/fabrics which connect di
- **High-memory nodes**: A mixture of 2x25 Gb Ethernet +RoCE and 100 Gb Ethernet
- **Scratch storage**: 25Gb Ethernet +RoCE

## Node Features
The Alpine cluster features some heterogeneity. A variety of feature tags are applied to nodes deployed in Alpine to allow jobs to target specific CPU, GPU, network, and storage requirements.

Use the `sinfo` command to determine the features that are available on any node in the cluster.

```bash
sinfo --format="%N | %f"
```

```{note}
**Feature descriptions and finalized partition names are still being added to Alpine nodes. Refer to the description of features list below for current node features.**
```

### Description of features
- **cpu**: AMD EPYC CPU (cores per node vary between 32-128)
- **a100**: NVIDIA A100 GPU
- **mi100**: AMD MI100 GPU
- **l40**: NVIDIA L40 GPU
- **h200**: NVIDIA H200 GPU
- **rtxpro6000**: NVIDIA RTX Pro 6000 GPU
- **storage**: large, fast RAID disk storage in node
- **rh8**: RedHat Enterprise Linux version 8 operating system
- **ib**: Features Infiniband, a high-speed message passing fabric for MPI jobs

## Job Scheduling

All jobs on Alpine are run through a queue system using the SLURM job scheduler. Though many HPC workflows are run through batch-type jobs, interactive jobs on compute nodes are allowed; however, these jobs must also be initiated through the scheduler. High-priority jobs move to the top of the queue and are thus guaranteed to start running within a few minutes, unless other high-priority jobs are already queued or running ahead of them. High-priority jobs can run for a maximum wall time of 24 hours. Low-priority jobs have a maximum wall time of 7 days.
Expand Down