Question: What is the Transition Timeline?
You will be transitioning from Wharton’s HPC3 cluster to PARCC’s Betty cluster from late summer to early fall 2026. Detailed timeline:
| Start Date |
Length | Details |
| 2026-08-12 (Wed) | 1 Month | a full set of HPC3 & PARCC resources will be available to your team during this transition period |
| 2026-09-16 (Wed) | 1 Month | 20% of a set of current HPC3 compute resources available during this transition period |
| 2026-10-14 (Wed) | 1 Month | HPC3 filesystem and two cores of HPC3 for data and code cleanup, archiving, migration, SAS Studio retired |
| 2026-11-18 (Wed) | 6 Month | Read only Globus access to data and code remains active, other services retired |
| 2026-06-17 (Wed) | 1 Month | Any remaining data and code archived to cold storage (if requested at that time) |
Question: What will change for Doctoral Students?
We will be adopting standard university doctoral student policies, requiring doctoral students to attain cluster access through a faculty sponsor, generally their advisor. We will assist with this process.
Question: How much will it cost to use the Cluster?
Base Fully-Subsidized by Wharton: as in our current Wharton cluster, Wharton will be fully-subsidizing a generous set of PARCC resources for your team’s use, which will be free for your team.
Compute: each PI has up to 128 cores of CPU and up to 4 NVIDIA B200 GPUs of concurrently-available compute resources.
Storage: each user has 50GB of their own, and each PI has a 1TB project folder. Those will be paid for by Wharton.
Pay-As-You-Go Resources: Above and beyond our free-for-you resources (listed above), you will be able to purchase more resources for your team. Rates are posted on the PARCC website here (use the Subsidized Rate column), and are payable by university budget code. Please have your faculty PI contact us at research-computing@wharton.upenn.edu with their Penn Budget Code and a list of Lab users who will be authorized to use these extended resources, and we will get you set up.
Question: Are there any HPC3 resources that will not be duplicated on PARCC?
Yes! There are several resources that will not be duplicated on PARCC:
- Mapped Network (SMB/CIFS) Drives: PARCC does not support SMB/CIFS (Windows Mapped Network Drive protocol). File transfers should be done using Globus or other standard SSH or SFTP-based methods.
- SAS Studio: Wharton’s SAS Studio service will be discontinued, but we will still offer SAS for all of our users. Access to graphical SAS will be possible through the PARCC Betty Open OnDemand service (graphical desktop and apps!).
- WRDS SAS data files (
/wrds): “mapping” WRDS data like this presents unacceptable risk to other Penn researchers on Betty. We recommend using WRDS data via their PostgreSQL service, and you can gain access to the WRDS cluster if you need access to these datasets in filesystem SAS format. WRDS connection methods are detailed here!
Question: What Data Classifications may be uploaded to and used on PARCC’s Betty Cluster?
PARCC’s Betty cluster is being built for use of Low and Moderate (non-PII, non-FERPA) data, as defined by Wharton’s Data Classification and Management Standard.
Question: Why do I see differences between storage usage and availability (Limit) reported when I log in?
Storage Limit values and rates are calculated in GB / TB (base-10 units). Depending on how you’re calculating usage, most OS tools use GiB / TiB (base-2 units). Read more here (includes a link to a calculator, too).
Question: For Wharton HPC3 users, is Transitioning to PARCC’s Betty Cluster optional?
No, it is not. Once we and our researchers agree that the current Wharton cluster is no longer needed, and after several months of maintenance without use, it will be retired. More below!
Question: How Will PARCC’s Betty Cluster Benefit Wharton Researchers, Wharton Computing, and Penn?
- Because of the personnel and financial resources that have and will be dedicated to PARCC, the quality, depth, and breadth of available cutting-edge services there far surpass what we are able to provide for our users. It is a “best of breed” system.
- These “economies of scale” also help greatly with PARCC’s vendor relationships. Support of the main products is direct, and high-touch.
- The billing-for-use model will allow for the team and the service to grow and keep pace with the latest in high-performance computing offerings available anywhere, at extremely competitive rates.
- These resources will be a great asset in attracting and maintaining Wharton’s world-renowned faculty and doctoral researchers.
- My team will be able to redirect a large portion of our current focus from “server administrators” towards “research collaboration”. We will be able to offer more extensive project assistance, whether that’s onboarding researchers and their RAs (training), assisting with code and workflow development, tuning, and modification for high-performance system use.
- While my team will no longer be directly “turning all of the knobs” on the systems, we will be responsible for managing resources for our researchers. We will also be working very closely with the team that is tasked with the physical and programmatic system work, as advocates for Wharton researchers in making the service a great experience.
- PARCC is the first pay-for-service HPC environment that will meet my team’s high standards and our researchers’ needs.
- Monies spent using PARCC services will stay within the Penn community, funding the future for the project.