Soonpil Kang's team at the Naval Postgraduate School chose a hard test case, Hurricane Guillermo, the ninth most intense Pacific hurricane on record.
The team had observational data from atmospheric scientist Stephen Guimond.
They needed a realistic, large scale tropical cyclone simulation that could be checked against it.
The work depends on fluid dynamics calculations, the kind GPUs are well suited to produce.
Using GPUs through NSF ACCESS allocations on NCSA's Delta supercomputer, the team ran its model five times faster than the CPU version.
Kang said GPUs could save both time and energy.
The goal is faster numerical weather prediction.
That is the payoff of affordable AI compute.
A small team gets more chances to run, check, and improve a serious model before the deadline arrives.
BHK Cloud gives US teams RTX 3090 GPU compute at $0.15/hr.
Buy Now Pay Later means zero upfront, payment after one month.
https://ai.bhkcloud.com/?utm_source=linkedin&utm_medium=social&utm_campaign=bhk_social_2026w39&utm_content=morning Which simulation would your team rerun this week if GPU time stopped slowing the work?
Originally posted on linkedin_personal