SCX_LAVD Core Compaction

SCX vs BORE - Cyberpunk

osu!stable + ffmpeg - BORE 5.2.6 - sched_burst_exclude_kthreads=1 vs sched_burst_exclude_kthreads=0

I ran benchmark with sched_burst_exclude_kthreads=0 twice after running one with sched_burst_exclude_kthreads=1 to make it effectively be BORE 5.2.6 vs BORE 5.2.5

osu!stable + ffmpeg - BORE v5.2.5 - Preserve 0x0 vs 0x42

Forza Horizon 5: All AMD GPU power modes (pp_power_profile_mode)

Basically 'echo "manual" > /sys/class/drm/card0/device/power_dpm_force_performance_level' and each option in /sys/class/drm/card0/device/pp_power_profile_mode

Forza Horizon 5: Benchmark different power modes of AMD GPU

Using CoreCTRL for this.

Forza Horizon 5: Benchmark consistency with identical settings

I've started a game and ran FH5 benchmark 3 times, one after another, on idle system, BORE scheduler, to understand about error size (e.g. how much each FPS avg can differ from other run). Number represent benchmark run index (again - identical settings, one after another).

CarX Drift Racing Online: BPFLAND gae86d2 vs gd61605

Twitch 1080p60 stream in the background Game all low 1080p

CarX Drift Racing Online: BPFLAND MAX_TASK_WAKEUPS

Twitch 1080p60 stream in the background Replay loop 80 seconds

Forza Horizon 5: Windows vs Linux

Using mesa-git on Linux, RADV driver. Can you tell why GPU power usage is much lower on Linux in comparison to Windows? Bug? :)