Della

Princeton's general-purpose research computing cluster, built up over several hardware generations and jointly funded by Astrophysics, the Lewis-Sigler Institute for Integrative Genomics, PICSciE and OIT. SDXCentral (an independent trade publication, successfully fetched) reported that Princeton added a roughly 300-strong NVIDIA H100 GPU cluster in March 2024, primarily financed by the university's endowment fund to support AI research including the Princeton Language and Intelligence (PLI) initiative; the same article independently confirms Della's existing NVIDIA A100 GPU fleet runs over FDR InfiniBand. Princeton's own systems/della page is the authoritative source for Della's full hardware table, but it returns HTTP 403 to automated fetches; per this dataset's sourcing policy that is blocked, not dead, so it is cited here as second_source_url even though this research could not read its content directly. A Princeton lab's public cluster-usage guide, hosted on GitHub and successfully fetched, corroborates a detailed node breakdown consistent with SDXCentral's report: AMD EPYC 9654 and Intel Cascade Lake CPU nodes, NVIDIA A100 GPU nodes (some MIG-partitioned), a 42-node / 336-GPU NVIDIA H100 'PLI' partition with a dedicated NDR InfiniBand fabric for GPU-to-GPU traffic, and an 18-node NVIDIA H200 'AI Lab' partition; that page is not an official Princeton IT channel, so it is not cited as this row's second_source_url, only described here for context. No aggregate FLOPS or total core figure was found on a page this research could fetch, so those fields, along with first_operational_date, stay blank rather than guessed.