GoldenCheetah memory consumption and recommendations - GoldenCheetah/GoldenCheetah GitHub Wiki

Resource Intensive Features

The ride cache precomputes all Metrics for all activities and for each interval and maintains them in RAM for faster access. Its not unusual to see GC consuming > 1GB of RAM. This isn't much different than other major desktop apps.

This does mean that startup and exit aren't instant since the ride cache must be loaded from disk on startup, and saved to disk on exit. The alternative would have been to compute these on demand but we designed GC to be nimble, at the cost of more "modern" hardware.

At present there are quite a few heavy computations:

367 metrics for every activity and every interval
17 mean maximal curves, 12 distribution curves, 7 time in zone distributions from every activity
6 pd model estimates for cp, w', cp/kg, w'/kg for rolling 6 week window over all history
identify and filter maximal performances across all rides

On top of this memory is consumed and kept:

activities you have viewed / edited since starting GC
chart curves for every chart for the current ride or date range selected
aggregated meanmax curves for every date range defined in Trends view

Memory requirements can be reduced by selectively disabling automatic interval discovery, user metrics or keeping historical data in a different athlete.

Minimum Hardware Recommendations

We would heartily recommend getting a SSD and replacing any old HD, and 8GB ram is pretty much the minimum for anyone these days and will be just fine unless you have > 1000 activities in your history.

Whilst we have seen GoldenCheetah running with some success on dual core i5 processors we would not recommend it. A dual core i7 CPU or equivalent is the minimum we would generally recommend.

We also strongly recommend a dedicated GPU such as NVidia or AMD to improve overall graphics rendering and performance, on board or software graphics are generally underpowered for the more complex charts in GoldenCheetah.