dynamic-occupancy-grid-map
dynamic-occupancy-grid-map copied to clipboard
Access Soa using AoS syntax
@cbachhuber can you see a runtime regression in comparison to the master branch? I'm seeing a small improvement which I didn't expect :smile:
Master:
DOGM cycle stats (14 splits):
Minimum: 62ms
Median: 70ms
Mean: 69.5ms
Maximum: 84ms
This branch:
DOGM cycle stats (14 splits):
Minimum: 66ms
Median: 70ms
Mean: 70.7143ms
Maximum: 85ms
So I don't really see an improvement :( For getting reliable, repeatable results, we should also think about longer runs. I think with 14 cycles, there is still too much variance involved to make meaningful judgements in the low one-digit millisecond range. Maybe you just made an improvement of 1-2ms, but we can't see that in the current data.
Definitely, we need like 100 iterations to get a real estimate. A real benchmark is needed to see those small improvements. But at least currently we can see that it's getting faster by the day even without real benchmarking.
So I don't really see an improvement :(
That's okay, I didn't expect one, that's why I was suprised because it was faster but maybe that just happened for that particular run.