English Amiga Board - View Single Post

Der Wanderer · 16 December 2013, 07:11

I am the author of MacroWave (not released yet).

I just want to confirm that I see some nice speedups with the native code, depending on the algorithm it is 1x up to 20x, I guess depends how much complex instructions like vectorisation can be used on the x86 side. This also indicates there is room for improvement for the JIT.

One thing I would like to see though is some kind of standard across emulations to embed x86 code (or say generally native code) into 68k.
E.g. a special 68K instruction like JNSR (jump native sub routine). Then we could make a shared library that contains native code. Right now it is a bit cumbersome to write a library that opens a DLL that must be installed separately on the Windows side/filesystem. It would be much better to be able to just replace a library with a mixed binary that contains the 68K stubs and the native code.

16 December 2013, 07:11	#6
Der Wanderer Registered User Join Date: Dec 2013 Location: Mountain View / USA Posts: 34	I am the author of MacroWave (not released yet). I just want to confirm that I see some nice speedups with the native code, depending on the algorithm it is 1x up to 20x, I guess depends how much complex instructions like vectorisation can be used on the x86 side. This also indicates there is room for improvement for the JIT. One thing I would like to see though is some kind of standard across emulations to embed x86 code (or say generally native code) into 68k. E.g. a special 68K instruction like JNSR (jump native sub routine). Then we could make a shared library that contains native code. Right now it is a bit cumbersome to write a library that opens a DLL that must be installed separately on the Windows side/filesystem. It would be much better to be able to just replace a library with a mixed binary that contains the 68K stubs and the native code.