Helix/Hardware: Difference between revisions
< Helix
Jump to navigation
Jump to search
S Richling (talk | contribs) |
S Richling (talk | contribs) |
||
| (28 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
== System Architecture == |
== System Architecture == |
||
The bwForCluster Helix is a high performance supercomputer with high speed interconnect |
The bwForCluster Helix is a high performance supercomputer with high speed interconnect. The system consists of compute nodes (CPU and GPU nodes), some infrastructure nodes for login and administration and a storage system. All components are connected via a fast Infiniband network. The login nodes are also connected to the Internet via Baden Württemberg's extended LAN BelWü. |
||
[[File:Helix-skizze.png| |
[[File:Helix-skizze.png|400px]] |
||
== Operating System and Software == |
== Operating System and Software == |
||
* Operating system: RedHat |
* Operating system: RedHat |
||
* Queuing system: Slurm |
* Queuing system: [[Helix/Slurm | Slurm]] |
||
* Access to application software: [[Software_Modules|Environment Modules]] |
* Access to application software: [[Software_Modules|Environment Modules]] |
||
| Line 27: | Line 27: | ||
! style="width:40%" colspan="3" style="text-align:center" | GPU Nodes |
! style="width:40%" colspan="3" style="text-align:center" | GPU Nodes |
||
|- |
|- |
||
!scope="column"| Node Type |
!scope="column"; style="background-color:#A2C4EB" | Node Type |
||
!style="text-align:left;"| cpu |
|||
| cpu |
|||
!style="text-align:left;"| fat |
|||
| fat |
|||
!style="text-align:left;"| gpu4 |
|||
!style="text-align:left;"| gpu4 |
|||
| gpu8 |
|||
!style="text-align:left;"| gpu8 |
|||
|- |
|- |
||
!scope="column"| Quantity |
!scope="column"| Quantity |
||
| |
| 355 |
||
| 15 |
| 15 |
||
| 29 |
| 29 |
||
| 26 |
| 26 |
||
| |
| 4 |
||
|- |
|- |
||
!scope="column" | Installed Working Memory (GB) |
!scope="column" | Installed Working Memory (GB) |
||
| Line 47: | Line 48: | ||
| 2048 |
| 2048 |
||
|- |
|- |
||
!scope="column" | Available Memory for Jobs (GB) |
!scope="column"; style="background-color:#A2C4EB" | Available Memory for Jobs (GB) |
||
!style="text-align:left;"| 236 |
|||
| 236 |
|||
!style="text-align:left;"| 2000 |
|||
| 2010 |
|||
!style="text-align:left;"| 236 |
|||
| 236 |
|||
!style="text-align:left;"| 236 |
|||
| 236 |
|||
!style="text-align:left;"| 2000 |
|||
| 2010 |
|||
|- |
|- |
||
!scope="column" | Interconnect |
!scope="column" | Interconnect |
||
| Line 68: | Line 69: | ||
| 8x [https://www.nvidia.com/en-us/data-center/a100/ Nvidia A100] (80 GB) |
| 8x [https://www.nvidia.com/en-us/data-center/a100/ Nvidia A100] (80 GB) |
||
|- |
|- |
||
!scope="column" | Number of GPUs |
!scope="column"; style="background-color:#A2C4EB" | Number of GPUs |
||
!style="text-align:left;"| - |
|||
| - |
|||
!style="text-align:left;"| - |
|||
| - |
|||
!style="text-align:left;"| 4 |
|||
| 4 |
|||
!style="text-align:left;"| 4 |
|||
| 4 |
|||
!style="text-align:left;"| 8 |
|||
| 8 |
|||
|- |
|- |
||
!scope="column" | GPU Type |
!scope="column"; style="background-color:#A2C4EB" | GPU Type |
||
!style="text-align:left;"| - |
|||
| - |
|||
!style="text-align:left;"| - |
|||
| - |
|||
!style="text-align:left;"| A40 |
|||
| A40 |
|||
!style="text-align:left;"| A100 |
|||
| A100 |
|||
!style="text-align:left;"| A100 |
|||
| A100 |
|||
|} |
|} |
||
The lines marked in blue are in particular relevant for [https://wiki.bwhpc.de/e/Helix/Slurm Slurm jobs]. |
|||
=== Intel Nodes === |
=== Intel Nodes === |
||
| Line 223: | Line 226: | ||
== Storage Architecture == |
== Storage Architecture == |
||
There is one storage system providing a large parallel file system based on IBM |
There is one storage system providing a large parallel file system based on IBM Storage Scale for $HOME, for workspaces, and for temporary job data. |
||
== Network == |
== Network == |
||
The components of the cluster are connected via two independent networks, a management network (Ethernet and IPMI) and an Infiniband fabric for MPI communication and storage access. |
The components of the cluster are connected via two independent networks, a management network (Ethernet and IPMI) and an Infiniband fabric for MPI communication and storage access. |
||
The Infiniband backbone is a fully non-blocking fabric with 200 Gb/s data speed. The compute nodes are connected with different data speeds according to the node configuration. |
The Infiniband backbone is a fully non-blocking fabric with 200 Gb/s data speed. The compute nodes are connected with different data speeds according to the node configuration. |
||
The term HDR100 stands for 100 Gb/s and HDR200 for 200 Gb/s. |
|||
Latest revision as of 16:58, 19 November 2024
System Architecture
The bwForCluster Helix is a high performance supercomputer with high speed interconnect. The system consists of compute nodes (CPU and GPU nodes), some infrastructure nodes for login and administration and a storage system. All components are connected via a fast Infiniband network. The login nodes are also connected to the Internet via Baden Württemberg's extended LAN BelWü.
Operating System and Software
- Operating system: RedHat
- Queuing system: Slurm
- Access to application software: Environment Modules
Compute Nodes
AMD Nodes
Common features of all AMD nodes:
- Processors: 2 x AMD Milan EPYC 7513
- Processor Frequency: 2.6 GHz
- Number of Cores per Node: 64
- Local disk: None
| CPU Nodes | GPU Nodes | ||||
|---|---|---|---|---|---|
| Node Type | cpu | fat | gpu4 | gpu4 | gpu8 |
| Quantity | 355 | 15 | 29 | 26 | 4 |
| Installed Working Memory (GB) | 256 | 2048 | 256 | 256 | 2048 |
| Available Memory for Jobs (GB) | 236 | 2000 | 236 | 236 | 2000 |
| Interconnect | 1x HDR100 | 1x HDR100 | 2x HDR100 | 2x HDR200 | 4x HDR200 |
| Coprocessors | - | - | 4x Nvidia A40 (48 GB) | 4x Nvidia A100 (40 GB) | 8x Nvidia A100 (80 GB) |
| Number of GPUs | - | - | 4 | 4 | 8 |
| GPU Type | - | - | A40 | A100 | A100 |
The lines marked in blue are in particular relevant for Slurm jobs.
Intel Nodes
Some Intel nodes (Skylake and Cascade Lake) from the predecessor system will be integrated. Details will follow.
Storage Architecture
There is one storage system providing a large parallel file system based on IBM Storage Scale for $HOME, for workspaces, and for temporary job data.
Network
The components of the cluster are connected via two independent networks, a management network (Ethernet and IPMI) and an Infiniband fabric for MPI communication and storage access. The Infiniband backbone is a fully non-blocking fabric with 200 Gb/s data speed. The compute nodes are connected with different data speeds according to the node configuration. The term HDR100 stands for 100 Gb/s and HDR200 for 200 Gb/s.