By Malcolm Kavalsky-Oracle on Oct 23, 2014
I was working with a leading asset management vendor in the financial server sector who is using Python for a considerable amount of their software, a typical three-tier architecture, Database, Business Logic and User interface using Python as the main back-end language. Performance was critical for both latency (fast individual query response) and total throughput (being able to service a large amount of queries in parallel).
This was an opportunity to validate the quality of the scalability advantage of SPARC processors, with their large amount of cores and threads within a single chip.
In order to test the scalability of the SPARC processor in a Python environment, I decided to use the standard Python benchmark which is available in all the latest Python distributions. By running multiple benchmarks in parallel, I could then plot the scaling factor to see how linearly the total throughput would ramp up as more cores and threads were utilized.