Shaheen II Decommissioning date (final)
Compute & Storage: April 25th, 2024 at noon
Shaheen III Project proposals
Accounts on Shaheen III are created following the submission and review of a project proposal. The Principal Investigator, who must hold a faculty position, is required to submit a project proposal detailing the nature of the work to be undertaken and the resources required for the project. There are two distinct types of projects:
- Development project: provides access for system familiarisation, code porting, performance assessment, and other pre-production work. This form can be used only for KAUST PIs for assessing software not installed on Shaheen. Development projects will not be allocated significant computing resources.
- Production project proposal : can be allocated significant computing resources. The RCAC reviews each project based on not only the computational readiness reviews but also on scientific readiness reviews by faculty experts in the scientific field of the proposed project.
Shaheen III CPU usage is free of charge for KAUST PIs and their team members at KAUST. PIs and users shall consult the Terms and Conditions of Usage.
Please download the project form from the one of the links above, complete it and submit your project proposal to https://apply.hpc.kaust.edu.sa/new_project/. For any inquiries, please contact projects@hpc.kaust.edu.sa
Comparison table Shaheen II and Shaheen III.
Description | Shaheen II (Cray XC40) | Shaheen III CPU Cray EX Genoa |
Nodes | 2 x Haswell 16 Cores | 2 x Genoa 96 cores |
Memory | 128 GB | 384 GB |
Nodes | 6174 | 4608 |
Peak node | ~1.1 TF | ~8.9 |
Theoretical Peak | 7.2 PF | ~40PF |
Total cores | 197,568 | 884,736 |
Interconnect | Aries | Slingshot 11 |
File systems | 18 PB at 0.5 TB/s | 25 PB HDD, 6.9PB Flash |
Applying for a Shaheen III account
The KAUST Supercomputing Laboratory manages three user account entities: organisations, projects, and users. Every Shaheen user must be an official member of at least one project and every project must originate from an approved organisation. Getting an account on Shaheen III is a three-step process and is detailed below.
- For non-KAUST users, your organisation or department must submit the Organisational Access Application (OAA), establishing a relationship between your home organisation and the KAUST Supercomputing Laboratory (KSL). The form should be completed, printed, signed, scanned and emailed to help@hpc.kaust.edu.sa. Each external organisation must have submitted a completed Organisation Access Application before its account requests or project proposals can be considered.
- The Principal Investigator (PI) must submit a Project Proposal Form as described above.
Each user must submit an Individual Account Application (IAA), supplying identification information from which we can generate login credentials.
Once the account is created, user will be notified by email and should follow these instructions to login.
Steps 2 and 3 should be done through the KSL Apply website using KAUST portal (Active Directory) credentials. If you are an external user who has used Shaheen II and still have an active portal account, you may use these credentials to access the website. Otherwise, you should email help@hpc.kaust.edu.sa attaching a scan of your passport photo page in order to obtain your credentials.
Shaheen II end of life
Shaheen II had its first users in production in July 2015, and since then, about 8 Billion core hours were used across 700 projects . The date of its decommissioning is April 25th 2024.
Shaheen II Storage Migration
As you all know, Shaheen II has 2 main storage systems:
- Scratch: used for temporary data while running SLURM jobs.
- Project: used to store important data such as results obtained from SLURM jobs.
The Scratch filesystem is already 8 years old and will NOT be migrated to Shaheen III. Shaheen III will come with its own Scratch filesystem. Any data left on Shaheen II Scratch will NOT be migrated over since it should not be important data as mentioned above. If there is any important data on Shaheen II scratch, please do get in touch with us as soon as possible so we can all work together to copy the data somewhere safe. The Project filesystem is relatively new so it will be migrated over to Shaheen III. However, users shall keep only necessary and active data for the migration.
As a reminder KSL does not manage archiving of data, it is under KAUST IT mandate with DataWaha.
It is IMPORTANT to know that:
- Quotas on Shaheen II Project storage will NOT be extended/increased when we migrate to Shaheen III. Copying all the data you have on Scratch to Project is NOT feasible.
- Projects will have to be resubmitted for RCAC approval as new projects on Shaheen III.
Additionally, it is important to know that any software you have compiled will have to be rebuilt to support the new Shaheen III CPU architecture.
The software maintained by the KSL CS Team will be recompiled by them for Shaheen III.
Our planned roadmap for storage decommissioning and migration is as follows:
- Expected date of decommission for compute nodes and running jobs: April 25th, 2024. On this date:
- ONLY compute nodes will be powered off, NO storage will be powered off.
- Shaheen II Scratch storage will be available READ-ONLY to copy data off of it. You will NOT be able to:
- Run jobs on it.
- Nor copy data to it.
- Firm date to power off scratch file system (/scratch): April 25th, 2024.
- During this period, Shaheen II scratch will be available READ-ONLY to copy data off of it. You will NOT be able to:
- Run jobs on it.
- Nor copy data to it.
- After April 25th, any data left on Shaheen II /scratch will be lost.
- During this period, Shaheen II scratch will be available READ-ONLY to copy data off of it. You will NOT be able to:
- Project storage /project/kxxxx should be curated and cleaned up before migrating from Shaheen II to Shaheen III. All PIs should go through the data they have in the Project directory and:
- Remove any data that is unnecessary.
- Verify the data owned by accounts that are disabled and delete if unnecessary.