i was wondering....would it be possible or efficient to rewrite the dolphin core in pc assembly and just do the video rendering in D3d or opengl? i know asm can achieve some insane speedups. i have heard that the leak of bleemcast (psx emulator for dreamcast) runs full speed but does not have full compatibility. this is amazing considering the dreamcast is not that much higher of a system except for the anti-aliasing. well its just an idea anyway. i dont know if the author knows or wants to take the time to learn assembly.
Dolphin core idea
- Thread starter taloon
- Start date
Dolphin uses dynamic recompilation (compiling blocks of PowerPC to raw x86 machine code and caching them), which (when perfectly implemented) is much much faster than any ASM interpreter. Dolphin's dynarec isn't perfect yet. It can probably improve in the future, though don't expect Wind Waker realtime 
Oh and the dreamcast has about six times the CPU power and even more times the rendering power of the PSX, so it's not THAT surprising that it can emulate the playstation.
Oh and the dreamcast has about six times the CPU power and even more times the rendering power of the PSX, so it's not THAT surprising that it can emulate the playstation.
Vegetable
Banned
Plus, doesn't the ps have only 2mb of vram? Although it is nice to see people bringing thoughtful ideas to the forum, and not asking how they can run they're gc discs on the dvd drive. How do you do that by the way?
*edit* I wouldn't give too many ideas for the next release of Dolphin, because if I was making an emu and everyone at the forum was always posting "OMG CAN YOU PUT CUP HOLDERS IN IT!??" I would be pretty pissed because I didnt ask for any ideas and wouldn't feel like programming a lot to it right away, we should all just let them relax and go at they're own pace, and it will probably be done quicker.
*edit* I wouldn't give too many ideas for the next release of Dolphin, because if I was making an emu and everyone at the forum was always posting "OMG CAN YOU PUT CUP HOLDERS IN IT!??" I would be pretty pissed because I didnt ask for any ideas and wouldn't feel like programming a lot to it right away, we should all just let them relax and go at they're own pace, and it will probably be done quicker.
Last edited:
dynarec shouldnt mess up any compatability if done right.. that might be why dolphin works so well
dynarec and interpretation should in theory have the exact same result if done right (100% compatability, speaking theoreticly), its just a little harder to get it working together when doing dynarec.
Scha
dynarec and interpretation should in theory have the exact same result if done right (100% compatability, speaking theoreticly), its just a little harder to get it working together when doing dynarec.
Scha
You'd probably be able to eliminate a bit of processing, but it becomes much harder to manage, at least IMO anyway. Besides, what you're talking about is a performance issue. The Dolphin team are still working on getting basic functionality. Once the basics are stable, inclusive, and well tested, you can then start thinking about what parts you might want to rewrite or optimize. This might well include rewriting or hand-optimizing sections that get called many times in ASM, but this is all still a fair way off yet. There are a lot of other areas where the time could be better spent right now.
Moogly Guy
New member
taloon said:
He already replied to what you said, you just failed to understand it. Essentially, a dynarec core would be faster than any other core, regardless of whether it was written in C, BASIC, or even assembly.
Doomulation
?????????????????????????
Fact is, that the compiler nowadays does a good job on optimizing code for optimal assembly. You need to know good asm to optimize small parts of the code. Yet, this is most of the time not needed.
Those who brag to write in asm are over exaggerating, 'nuff said.
Those who brag to write in asm are over exaggerating, 'nuff said.