Helix/Hardware: Difference between revisions

From bwHPC Wiki
Jump to navigation Jump to search
 
(94 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. Is composed of login nodes, compute nodes and parallel storage systems connected by fast data networks. It is connected to the Internet via Baden Württemberg's extended LAN BelWü.
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|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]]


== Compute Nodes ==
== Compute Nodes ==


The cluster is equipped with the following CPU and GPU nodes.
=== AMD Nodes ===


⚫
{| class="wikitable" style="width:80%;"
Common features of all AMD nodes:
⚫
* Processors: 2 x AMD Milan EPYC 7513
⚫
* Processor Frequency: GHz
* Number of Cores per Node: 64
* Local disk space: None

⚫
{| class="wikitable"
|-
|-
! style="width:10%"|
! style="width:20%" |
! style="width:10%" colspan="2" style="text-align:center" | CPU Nodes
! style="width:40%" colspan="2" style="text-align:center" | CPU Nodes
! style="width:10%" colspan="3" style="text-align:center" | GPU Nodes
! style="width:40%" colspan="4" 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
| colspan="2" style="text-align:center" | gpu4
!style="text-align:left;"| gpu4
!style="text-align:left;"| gpu4
| gpu8
!style="text-align:left;"| gpu8
!style="text-align:left;"| gpu8 (in preparation)
|-
|-
!scope="column"| Quantity
!scope="column"| Quantity
| xxx
| 355
| xxx
| 15
| xxx
| 29
| xxx
| 26
| 2
| 4
⚫
| 3
⚫
|-
!scope="column" | Processors
⚫
| 2 x AMD EPYC 7513
| 2 x AMD EPYC 7513
| 2 x AMD EPYC 7513
| 2 x AMD EPYC 7513
| 2 x AMD EPYC 7513
| 2 x AMD EPYC 9334
⚫
|-
⚫
!scope="column" | Processor Frequency (GHz)
| 2.6
| 2.6
| 2.6
| 2.6
| 2.6
| 2.7
⚫
|-
!scope="column"; style="background-color:#A2C4EB" | Number of Cores per Node
!style="text-align:left;"| 64
!style="text-align:left;"| 64
!style="text-align:left;"| 64
!style="text-align:left;"| 64
!style="text-align:left;"| 64
!style="text-align:left;"| 64
|-
|-
!scope="column" | Working Memory (GB)
!scope="column" | Installed Working Memory (GB)
| 256
| 256
| 2048
| 2048
Line 44: Line 67:
| 256
| 256
| 2048
| 2048
| 2304
|-
!scope="column"; style="background-color:#A2C4EB" | Available Memory for Jobs (GB)
!style="text-align:left;"| 236
!style="text-align:left;"| 2000
!style="text-align:left;"| 236
!style="text-align:left;"| 236
!style="text-align:left;"| 2000
!style="text-align:left;"| 2200
|-
|-
!scope="column" | Interconnect
!scope="column" | Interconnect
Line 50: Line 82:
| 2x HDR100
| 2x HDR100
| 2x HDR200
| 2x HDR200
| 4x HDR200
| 4x HDR200
| 4x HDR200
|-
|-
Line 55: Line 88:
| -
| -
| -
| -
| 4x [https://www.nvidia.com/en-us/data-center/a40/ Nvidia A40]
| 4x [https://www.nvidia.com/en-us/data-center/a40/ Nvidia A40] (48 GB)
| 4x [https://www.nvidia.com/en-us/data-center/a100/ Nvidia A100]
| 4x [https://www.nvidia.com/en-us/data-center/a100/ Nvidia A100] (40 GB)
| 8x [https://www.nvidia.com/en-us/data-center/a100/ Nvidia A100]
| 8x [https://www.nvidia.com/en-us/data-center/a100/ Nvidia A100] (80 GB)
| 8x [https://www.nvidia.com/en-us/data-center/h200/ Nvidia H200] (141 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
!style="text-align:left;"| 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
!style="text-align:left;"| H200
|-
!scope="column"; style="background-color:#A2C4EB" | GPU Memory per GPU (GB)
!style="text-align:left;"| -
!style="text-align:left;"| -
!style="text-align:left;"| 48
!style="text-align:left;"| 40
!style="text-align:left;"| 80
!style="text-align:left;"| 141
|-
!scope="column"; style="background-color:#A2C4EB" | GPU with FP64 capability
!style="text-align:left;"| -
!style="text-align:left;"| -
!style="text-align:left;"| no
!style="text-align:left;"| yes
!style="text-align:left;"| yes
!style="text-align:left;"| yes
|}
|}


The lines marked in blue are in particular relevant for [https://wiki.bwhpc.de/e/Helix/Slurm Slurm jobs].
=== Intel Nodes ===

Some Intel nodes (Skylake and Cascade Lake) from the predecessor system will be integrated. Details will follow.


== Storage Architecture ==
== Storage Architecture ==


There is one storage system providing a large parallel file system based on IBM Spectrum Scale for $HOME, for workspaces, and for temporary job data.
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. The compute nodes do not have local disks.


== 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 requirements of the 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 21:59, 15 April 2025

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.

Helix-skizze.png

Operating System and Software

Compute Nodes

The cluster is equipped with the following CPU and GPU nodes.

CPU Nodes GPU Nodes
Node Type cpu fat gpu4 gpu4 gpu8 gpu8 (in preparation)
Quantity 355 15 29 26 4 3
Processors 2 x AMD EPYC 7513 2 x AMD EPYC 7513 2 x AMD EPYC 7513 2 x AMD EPYC 7513 2 x AMD EPYC 7513 2 x AMD EPYC 9334
Processor Frequency (GHz) 2.6 2.6 2.6 2.6 2.6 2.7
Number of Cores per Node 64 64 64 64 64 64
Installed Working Memory (GB) 256 2048 256 256 2048 2304
Available Memory for Jobs (GB) 236 2000 236 236 2000 2200
Interconnect 1x HDR100 1x HDR100 2x HDR100 2x HDR200 4x HDR200 4x HDR200
Coprocessors - - 4x Nvidia A40 (48 GB) 4x Nvidia A100 (40 GB) 8x Nvidia A100 (80 GB) 8x Nvidia H200 (141 GB)
Number of GPUs - - 4 4 8 8
GPU Type - - A40 A100 A100 H200
GPU Memory per GPU (GB) - - 48 40 80 141
GPU with FP64 capability - - no yes yes yes

The lines marked in blue are in particular relevant for Slurm jobs.

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. The compute nodes do not have local disks.

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.

Retrieved from "https://wiki.bwhpc.de/wiki/index.php?title=Helix/Hardware&oldid=14741"