For any dynamic game, especially multiplayer. except for strategies, any fps below 60 in "high" quality can be considered a priori unplayable process.
And again I will repeat once again - the main problem of igpu is slow RAM, which is also used by the system, drivers and software, in addition to igpu.
Until the SoC crystal has dedicated VRAM with a bandwidth of more than 300 GB / s, there will be no serious progress in igpu, no matter how many shader units they add, as there will be no serious improvement in multi-threaded performance in real software when increasing the processor cores for exactly the same reason.
The bandwidth of x86 RAM lags behind the hardware requirements by several times, which is why Intel / AMD are desperately increasing processor caches that are meaningless under serious loads. This is a poultice for a dead man.
Until there is a sharp improvement with the x86 memory subsystem, progress will be at a snail's pace against the background of dGPU. Just like Apple, where theoretical 400 GB/s is declared in top configs, and in reality a shameful 120 GB/s in Max versions, which gives a completely shameful efficiency of the memory controller.
With real 400 GB/s, Apple's igpu chips would have confidently caught up with 4060+ long ago, but in reality the performance is much worse, which once again proves where the real bottleneck of all igpu is.