This reduces code size more than 20% (which is 200-300KB!).
DMA handling accounts for less than 0.5% the average emulation runtime
which doesn't justify the crazy optimization level that the code has. In
fact it's more than likely that the new code runs faster due to less
I-cache trashing.
This reduces code size more than 20% (which is 200-300KB!). DMA handling accounts for less than 0.5% the average emulation runtime which doesn't justify the crazy optimization level that the code has. In fact it's more than likely that the new code runs faster due to less I-cache trashing.