search

FLAMEGPU / FLAMEGPU2

FLAME GPU 2 is a GPU accelerated agent based modelling framework for CUDA C++ and Python
https://flamegpu.com
MIT License
106 stars 22 forks source link

BugFix: CUDASimulation nolonger reports telemetry incorrectly... #1081

Closed Robadob closed 1 year ago

Robadob commented 1 year ago

when an submodel or ensemble run.

config.is_ensemble is set by both ensemble and submodel runs. Closes #1079

Robadob commented 1 year ago

Unfortuantley this has conflicts now with something else that has been merged.

Will rebase

I'm not sure that reporting telemetry per simulation in an ensemble is useful, other than making numbers bigger?.

That's an argument to be had with @mondus

Robadob commented 1 year ago

Could rework this further and remove is_ensemble from CUDAConfig, and check based on ensemble_run_id instead. Possible also of value to create an enum/const for UINT_MAX, bit easier for Python users to check against.

ptheywood commented 1 year ago

Merging as desired by @mondus. Independently telemetry payloads for ensmble runs now include the ensemble size and concurrent batch size (#1098)