CarX Drift Racing Online: BPFLAND slice comparision

Youtube 720p music playlist in the background + Discord Call + Game in 1080p all low

Mirror's Edge: bpfland-next commits, under stress -c $((3 * $(nproc)))

Last daily commits of past several days, in bpfland-next branch. Under stress -c $((3 * $(nproc))) command.

Mirror's Edge: bpfland-next commits, idle system

Last daily commits of past several days, in bpfland-next branch. Under idle system.

CarX Drift Racing Online: BPFLAND slice comparision

1080p all low

CarX Drift Racing Online: BPFLAND vs BORE

This time with no background load, only game was running

CarX Drift Racing Online: BPFLAND vs BORE

The following command was running in the background: stress-ng --cpu-method loop -c $(($(nproc)*4)) 5000 vs 20000 max slice

CarX Drift Racing Online: BPFLAND vs BORE

This game is running a replay loop of the same race, so it never changes Game resolution set at 720p all low to increase cpu usage Twitch stream 1080p60 running on the background also tried different maximum slices for bpfland

scx_bpfland bruteforce nvcsw-max-thresh parameter (bpfland-next, f80123c)

I've noticed in https://flightlesssomething.duckdns.org/benchmark/22 that --nvcsw-max-thresh 60 gave much higher FPS than the rest, so decided to double check this...

scx_bpfland bruteforce of each parameter (bpfland-next, f80123c)

scx_bpfland bruteforce of each parameter (bpfland-next, f80123c)

Forza Horizon 5 - Windows vs Linux performance

Windows 11 vs CachyOS w/ scx_rusty. Metrics are taken from Afterbuner/RivaTuner and modified by hand to match MangoHud format, so CSVs can be uploaded for comparison.