> Funny thing is that the G5 problems were already mentioned:
> http://euaejit.blogspot.com/2012/10/jinxed-it-more-apples.html :)

Indeed, and the solution has already been worked out back then:

"Tobias Netzel [...] managed to fix up E-UAE for the G5 (PowerPC 970) processors. He got rid of the mcrxr emulated unsupported instruction [...] for this processor type. He also done some optimizations regarding the microcoded instruction usage [...]. [...] almost half of the emulated instructions make use of the mcrxr instruction for emulating some arithmetic flags. With the recent changes this PPC instruction is not used for anything if the emulator was compiled for G5 and the fix helps a lot on the interpretive emulation too."

To me as a non-developer, this sounds like it just has to be compiled for PPC970 target and the alternative G5-optimized code should be used automatically for the resulting binary.


Edit:

"Performance of UAE used to be bad on G5 CPUs. The reason for this is the use of the "mcrxr" instruction which CPUs of POWER4 and later POWER architecture versions don't implement at all. For that reason OSes like GNU/Linux or Mac OS X/Darwin catch the illegal instruction exception that is thrown when that instruction is being executed and emulate it in software. That results in a huge performance penalty. That "mcrxr" instruction is used explicitely in the optimized flags code. Turning off flags optimization when executing gencpu will restore performance on those CPUs. Additionally I contributed code to FS-UAE and E-UAE PowerPC JIT that works around that problem."
http://www.freelists.org/post/uae/Performance-on-G5-or-POWER4-and-later-CPUs

> To me as a non-developer, this sounds like it just has to be compiled for PPC970 target and the alternative G5-optimized code should be used automatically for the resulting binary.

Yes, Andreas, this is what it should do. For this, there is a "define" to enable it. Unfortunately I don't have a G5 computer to test it, but now I found some ppl who will. So the Beta1 is strongly focused on the JIT implementation.
But don't worry, there will be a G5 version with JIT.

> there will be a G5 version with JIT.

Now there:

"thanks to Fab the G5 executable is also available in the release package. [...] there were some murmur about the speed of the MorphOS version on G5 on some forums. No wonder: it needs a special version, which can be compiled from the sources for some time now. (Thanks to Tobias Netzel and to Fab for the special build.) Probably the same applies to the PA Semi processor"
http://euaejit.blogspot.com/2014/03/ppcjitbeta02-beginning-of-beautiful.html

Cool news Andreas, thank you!
Downloaded and tested a bit.... Fast, really fast on a G5 2.7 and it is also stable.
Very good work.

I have tried, the 2nd beta on my 1.33ghz mac mini, but i have many sound interrupts when i want to play a game or a demo,
The first version from fab, whitout jit has not this problemes especially on games. For demos like from elude the jit version seems to be twice faster in framerate but the sound is desapointing

Update:

>> there will be a G5 version with JIT.

> Now there: [...]

The new beta doesn't come with a G5 built for MorphOS, unfortunately:

"as it seems there is no official compiler with G5 support yet in the MorphOS SDK and it is rather complicated and unreliable to compile any source for that processor. Until this situation is improved the G5 version for MorphOS won't be available from the beta binaries."
http://euaejit.blogspot.com/2014/05/ppcjitbeta03-switch-to-ludicrous-speed.html

"The compiling for G5 is still not resolved for MorphOS, no G5-optimized binary again, sorry guys. If you could tell me how can I (easily) compile the files for G5 on my iBook then I give it a go, I promise."
http://euaejit.blogspot.com/2014/07/ppcjitbeta04-fourfourfourfour.html

Fab?

Edit1: Fab did it
Edit2: added beta 4 quote
Edit3: Fab did it again
Edit4: And once more

[ Edited by Andreas_Wolf 11.01.2015 - 14:28 ]

I am curios. Does Fab have his own gcc on Morphos or is he using cross compiling?

> Does Fab have his own gcc on Morphos or is he using cross compiling?

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=10143&forum=9&start=1