Philosophical question related to AROS vs OS4

[bubblebobble]

I think it is a myth that OS friendly 68K programs are executed faster on OS4 or MOS, because their emulation is more lightweight.

If you are not playing games, but running a "High-End" Amiga, you don't need cycle exact emulation of Custom Chips. Actually, apart from CIA, you don't need any Custom Chip emulation if you are using an RTG Screen and AHI Sound, and WinUAE is doing exactly that.

So what has to emulated is redicously cheap to emulate and costs a neglegtable amount of CPU Power. The execution of 68K programs under WinUAE has NO handbrake on, as people post here.

[bubblebobble]

I was talking about the 68K Emu, not how well the RTG Driver supports the hardware.
WinUAE has great room for improvement here, indeed, since it is pure software as far as I know.
But if you run the test on actual 68K code, the results will look very different. See the FFT Benchmark which I wrote and Bernd send around. You can also check DSP effects of HD-Rec.

Or AmegaOne (68K), this runs much faster on WinUAE than on OS4/MOS, how come?

All I want to say is that WinUAE execution speed of 68K Code is faster than OS4/MOS. I am not comparing Gaphicscard drivers.

[bubblebobble]

Funny you're comparing 68k Emu and you are showing something that involves (RTG) graphics, and not really 68k... RTG Graphics stuff *is* much faster on WinUAE...

Basic CPU emulation stuff is slower on WinUAE.Now I don't know why, you should ask some developer for that. But it's a fact.

Well, hard to show something that does not involve (RTG) graphics. Given that it is rendered in software, I count it as CPU intensive.
(BTW, OT: Does SDL.library use HW acceleration on OS4/MOS ?)

But I also mentioned Bernds FFT Benchmark (FPU and non-FPU version) and some HD-Rec DSP Effects (Interger and or float).