HPC Questions
HPC Questions
1 What is Raad2 ?
Raad2 is our current flagship system.
It is a Cray XC40 with 4,128 Intel Xeon Haswell cores, running under Linux SLES distribution.
Put into production in early 2017, it has 172 compute nodes interacting via an Aries dragonfly interconnect, with an aggregate peak performance of 120+ TFLOPS(Linpack).
Each compute node contains 2 sockets with 12 physical cores per processor chip -- giving a total of 24 cores per node, along with 128GB of RAM.
More detailed can be found here: Raad2 intro
Raad2 system also has a separate GPU Cluster equipped with NVIDIA V100 GPUs and Intel Xeon Skylake processors.
Users who want to accelerate their AI, HPC or Data science applications can largely benefit from this resource.
More detailed can be found here: Raad2-gpu System_Overview
2 What is Hazeem ?
Hazeem is another HPC Cluster that has been operated by QEERI (HBKU) for several years.
Its management has recently been handed over to our RCCG Team, and it has been opened to any HBKU member with a computationally intensive need that can be justified by a PI.
Architecturally, it is very similar to Raad2 system. It is in fact a CRAY XC50 product, successor to the CRAY XC40 (Raad2).
Concisely, Hazeem system specifications include:
- Compute nodes: 120
- CPU cores & Memory per node: 40 cores, 192 GB
- CPU Type: Intel Xeon Gold 6142 (Broadwell)
- Aggregate CPU Cores: 4400+
- Aggregate System Memory: 22TB+
- Interconnect: Aries Dragonfly
- GPUs: None
3 How to access HBKU HPC systems ?
To access our HBKU HPC Clusters you first need to create a HBKU HPC Account : User accounts
Once your account is approved, you will receive an email notification containing temporary login credentials and some basic guidance on how to access the system.
You can also refer to this page for additional information: Raad2 getstarted
4 What applications are available on the HBKU HPC Clusters?
Our team maintains a set of applications that are mainly targeting scientific computing domain.
To know the full list of applications installed on our HPC Clusters, please refer to this page: Application Guides
5 How to launch my program/algorithm/script in the HPC Clusters?
Once you established the connection to Raad2, you are connected to a "login node".
To launch a computationally intensive program/algorithm/script, also called a "job", you need to use a program called SLURM.
SLURM will serve as the bridge towards the multiple "compute nodes" of the system.
These are the nodes designed to handle computational workload efficiently.
To learn how to use SLURM to launch different kinds of job, please refer to Raad2 Submitting Jobs
If you want to go further, please check Types of Jobs
6 How to transfer files from my PC to the HPC Clusters and vice versa ?
There are different ways to achieve this task, but we will explain two of them.
Our recommended approach is to use a (free) utility called WinSCP, for more details please refer to: WinSCP
The alternative is to use the transfer built-in features of MobaXterm.
7 How to access Raad2 GPU Cluster ?
To access our Raad2 GPU Cluster you first need to create a HBKU HPC Account : User accounts
Don't forget to select "yes" in the field "GPU's Required?"
Once your account is approved, you will receive an email notification containing temporary login credentials and some basic guidance on how to access the system.
You can also refer to this page for additional information: Raad2-gpu Getting_Started