mirror of https://github.com/LTTLabsOSS/markbench-tests.git synced 2026-01-14 00:18:11 -05:00

Files

j-lin-lmg 36c542be4a James round seconds (#129 )

this is a ford of jd/harness-fixes that is currently live on the
benches, the ONLY thing i changed is im rounding the timestamps to the
nearest second

please approve

---------

Co-authored-by: J-Doiron <139803019+J-Doiron@users.noreply.github.com>

2025-04-09 14:34:19 -07:00

settings

initial stellaris test:

2024-08-29 14:46:53 -07:00

manifest.yaml

initial stellaris test:

2024-08-29 14:46:53 -07:00

README.md

update

2024-08-30 10:06:41 -07:00

stellaris_utils.py

missed a function call

2024-08-30 13:46:10 -07:00

stellaris.py

James round seconds (#129 )

2025-04-09 14:34:19 -07:00

README.md

Stellaris

This benchmark uses vanilla Stellaris (No DLC) and the popular console command: one_year. We created a save game:

30 advanced start AI empires
Max saturation of hyperlanes
Maximum galaxy size

We used the fast_forward command to get to year 2400.

Prerequisites

Python 3.10+
Stellaris installed
Keras OCR service

Options

kerasHost: string representing the IP address of the Keras service. e.x. 0.0.0.0
kerasPort: string representing the port of the Keras service. e.x. 8080

Output

report.json

resolution: string representing the resolution the test was run at, formatted as "[width]x[height]", e.x. 1920x1080
start_time: number representing a timestamp of the test's start time in milliseconds
end_time: number representing a timestamp of the test's end time in milliseconds
score number of seconds to advance one full year in game