Dreamshell RC3 vs GDEMU

[SWAT]

Thanks! Really useful to see a comparison.
I think I can increase the loading speed for the IDE DMA, at the moment it works with some delays (not lower speed in general) because polling state is done every frame, rather than IDE interrupt. The smaller the size of the cluster in FAT, the more delay occurs.
I'll try to fix it in the new version, I'm just wondering, maybe can I do faster than GDEMU Although in reality it is not necessary, just sporting interest