Why is ik_llama.cpp said to be faster than Mainline? On my hybrid multi-GPU rig, Mainline easily beats it

I constantly see recommendations saying that ik_llama.cpp (ikawrakow's fork) is the undisputed king of hybrid CPU/GPU offloading and MoE performance. However, every time I benchmark it against mainline ggml-org, mainline consistently beats it by a wide margin. Am I missing specific flags, or is ik_l