M68K for DC

[Eric]

Does this mean that Stef would leave the scene if we have you working in the scene. This is something thats been on my mind. Although in the real world if sh4 asm is way better then C++ and if that was how Sega was working i know they werent cause they were more complex with Katana but if they were they would probably go with a sh4 asm before C++.

I am not trying to pee off anyone but in the real world thats what could happen or even Stef could work with this guy and see what kinda ideas can run through to make althese emulators run quickly.

Am I right?

Eric

[BlackAura]

[quote author=Eric link=board=dcemu;num=1083323639;start=285#290 date=08/01/04 at 19:28:29]Does this mean that Stef would leave the scene if we have you working in the scene. This is something thats been on my mind. Although in the real world if sh4 asm is way better then C++ and if that was how Sega was working i know they werent cause they were more complex with Katana but if they were they would probably go with a sh4 asm before C++. Eric[/quote]

I wouldn't think so. There's still lots of work to go around, and C68k did make a massive difference over Musashi.

Anyway, adding it to Genesis Plus shouldn't be that hard, becauase when Stef added it, he created a wrapper module which can be used to change the CPU core over fairly easily. At the moment, for the hardware accelerated version, we don't need that much extra speed. We're probably around 95% with sound, so we only need another ~5% somewhere.

[Eric]

Would people see a difference with that 5%?

[Alexvrb]

[quote author=Eric link=board=dcemu;num=1083323639;start=285#288 date=08/01/04 at 19:02:14]Okay what about SNES emulators then?[/quote]
No. You must not have read through the early stuff here. C68k, m68k, Fox's 68k, or any other 68k emulator all emulate the Motorola 68k. That CPU is used in Genesis, NeoGeo CD, and a good amount of arcade systems. If a system doesn't have a 68k chip in it, the emulator for 68k isn't going to help. SNES, PSX, N64, etc, don't have 68k chips. So you aren't going to be using C68k or any other 68k emulator for them. You could always use www.google.com and find out more information about what hardware these systems use on your own.

Saturn actually *does* have a 68k in it, but it's part of the sound subsystem, not the main processor. I'm not even sure how important accurate emulation of it is, so you still might not need a full 68k emulator. Plus it wouldn't really help since the rest of the system needs to be emulated properly first.

[quote author=Eric link=board=dcemu;num=1083323639;start=285#292 date=08/01/04 at 19:59:12]Would people see a difference with that 5%?[/quote]That depends. If your sound is breaking because the emulator is too slow, then the 5% would fix the sound. You might not notice the speed difference, but the sound wouldn't screw up on you anymore, and that is a BIG difference. Especially when you're working to retain authentic Genesis audio. Which is why Quzar told Stef that he'd be happy with a Z80 that is only a little faster, as both emulators would benefit from the speed - and probably from a more properly working Z80 emu, too.

Fox: What are you gonna call it? F68k? Â* 8)

[quzar]

[quote author=Ian_micheal link=board=dcemu;num=1083323639;start=285#289 date=08/01/04 at 19:20:07]


Hmm if you did you never gave it to me working i worked with stefd he put the core in and to compile it needed precompiled files and 1 precompiled from me.

HLE stuff was not added when you attemped it no way it would work.

I dont want stef to read that becuase he put the core in neogeo cd and showed me the functions after that i understood it was able to get it to compile then helped him to get it to compile with some little setup problems in neogeo cd.


Stef put the core in it sent me the files which still would not compile due to some strange problems which i fixed.

I can show you how and were and what to do with the new core.

Sorry if this look mean or a reproch but stef d Â* did it not you.

Dont mean any disrespect but if i was stef and seen your comment it would slightly piss me off.
[/quote]
After stef gave it to you you asked me to replace all the memory functions because it wasnt compiling for you. He compiled the core and you gave me that, then i changed all the cpu calls to the different ones and got the cpu68k working with musashi correctly. If you recall there were issues because of the directory structure and where all the files were.

That dosnt matter though, in any case I am also capable of implementing it.

I want to see what stef would have to ask fox as to the design of the cpu core.

@Fox: I truely mean no disrespect from this, but what have you done before? Have you ever written 68k cores or something? I would reccomend looking to see how Stef's core works and keeping that in mind as you finish this, because just asm wont give miracle speedups.

[quzar]

[quote author=Alexvrb link=board=dcemu;num=1083323639;start=285#293 date=08/01/04 at 20:16:26]Saturn actually *does* have a 68k in it, but it's part of the sound subsystem, not the main processor. I'm not even sure how important accurate emulation of it is, so you still might not need a full 68k emulator. Plus it wouldn't really help since the rest of the system needs to be emulated properly first.[/quote]

Its not a standard 68000, i think its a 68020 or 68E20 or something like that, ATM I don't recall the exact lettering. But I'm pretty sure that would only be a small issue as the whole 68xxx line is very similar.

[guymelef]

no bickering it could only lead to ruin.
teamwork can get this thing of the ground without a hitch in a year easy

[Stef]

[quote author=quzar link=board=dcemu;num=1083323639;start=270#282 date=08/01/04 at 14:24:41]stefs 68k core is in C, but it is very very well written. He says that by the way it works it would be hard to get more speed by using asm. There were still a few things he wanted to do to gain a bit more speed.[/quote]

I don't remember where i said that ... There is always way to do better in ASM, specially with GCC and SH-4 code

Here's what i said in a old post :

Then about SH-4 assembly versus C generated code.
First GCC isn't really a master about SH-X code optimisation, it does some weird stuff sometime and code is far from being optimal.
Second, when you're emulating a CPU, asm always help a lot compared to C, because you have access to some low level instruction as ROR / ROL which need several instruction in C.
So, even with the best C code, i think you can at least do 3 time faster in ASM.

C68K is definitly not the ultimate 68k emulator for the DC, we can do really better in ASM, but, that's really needed ? if C68K is already capable of emulating a 68000 running at 80/100 Mhz on the DC, then an ASM 68000 emulator isn't really a must.

So we can definitly do better with ASM, that's why fox68k's new SH-4 assembly core is welcome
You can easily test it in Genesis Plus DC or NeoCD by implementing it in the CPU interface (cpu68k.c file).
But as your core isn't very advanced (lack of interrupt) you should test it with a small test code and compore it against others cores
Are the sources availables somewhere ? or you want to first complete it before releasing anything ?

[quzar]

Im sorry i thought taht at some point you said something along the lines of "it would still take a well written asm core to match this one" or something like that... sorry :-/ i think i should take a break from posting for a bit.. sigh.

[Stef]

[quote author=quzar link=board=dcemu;num=1083323639;start=285#298 date=08/02/04 at 03:02:55]Im sorry i thought taht at some point you said something along the lines of "it would still take a well written asm core to match this one" or something like that... sorry Â*:-/ i think i should take a break from posting for a bit.. sigh.[/quote]

Yeah i remember, i said that... a bad ASM core can be slower than a good C core, but i guess fox68k core has been disigned for speed And a good C core can't be as fast than a good ASM core, that's the idea