Data processing
Dataset ingestion, joins and repeated scans.
Read/write throughput and processing time.
Hardware partnerships
PSIM seeks a sponsored rack server in exchange for documented benchmarks and a public case study.
The configuration is flexible, with storage, memory and CPU capacity as the priorities.
Research workloads
Feature construction
Larger working sets and repeated transformations.
Peak memory, elapsed time and dataset-size scaling.
Repeated evaluations
Walk-forward tests, cost sensitivity and concurrent jobs.
Job throughput, runtime and CPU use.
Dataset sizes and concurrency would be matched to the supplied system. Benchmarks would document software, inputs, run conditions and limitations.
Practical details
Before testing, we would agree the configuration, ownership or loan terms, installation, milestones and publication rights. Licensed datasets remain private.
The case study would be published on psim.one and could be used on the partner’s website or technical blog, with attribution and factual review agreed in advance.
PSIM is an independent research project by Adam Rosta, based in Switzerland.