Printable View
pretty much everything is working well with the exception of too much flash rom being used.
mabey some more speed will come next... i hope.
The splash (thanks to Neo_Matrix_IV :) ) looks very nice on the tv screen. and the sound is vastly improved... no more of that damn ticking. things seem to be moving smoothly and its nice to have GPF back. ;D
i havent seen this emulator yet, but from the sounds of it, it seems pretty cool....i cant wait till its completed :)
Nice to hear things are improving :)
remember that if u need help with some images, the only thing you have to do is ask ;)
[quote author=Neo_Matrix_IV link=board=dcemu;num=1098286979;start=60#62 date=12/03/04 at 13:23:05]Nice to hear things are improving :)
remember that if u need help with some images, the only thing you have to do is ask ;)[/quote]
Will do.
Thanks again for the awesome splash screen.
Troy
Hi folks!
I'm currently porting NeoPop to a SuperH2 CPU architecture.
There where a couple of problems involved, because the TLCS900h CPU core of NeoPop is only for LSB (like x86) architecture and got no (unlike the sound CPU z80 core) MSB compile flag.
Yes, I've tried very hard to fix this alone, but now help would be very appreciated.
I got it already running, but had only graphics with one homebrew demo. Some tiles are missing and it was really hard to find the reason for that. Now that my fix still doesn't work, I really need help.
All accesses to memory with _u16 and _u32 pointers are converted to return MSB value format and furthermore proper alignment is established.
Now my last problem - I thought - has something to do with incorrect register access.
Remember these bunch of mapB mapW and mapL files in the TLCS900 core?
Looking into these I saw accesses to 32bit register values with word and byte pointers, expecting to have it in LSB format. Of course the LSB access to the first byte of a 4 byte MSB register will return the last byte instead.
So I decided to modify these .h files to be compilable for both, MSB and LSB.
Now please, can you give some hints to me, what fixes you established regarding alignment and endianes?
Thanks,
Rockin'-B
That's great news, Rockin'-B! Wish you all the best & welcome to DCEmuUK :)
[quote author=RockinB link=board=dcemu;num=1098286979;start=60#64 date=01/02/05 at 12:29:02]Hi folks!
I'm currently porting NeoPop to a SuperH2 CPU architecture.
There where a couple of problems involved, because the TLCS900h CPU core of NeoPop is only for LSB (like x86) architecture and got no (unlike the sound CPU z80 core) MSB compile flag.
Yes, I've tried very hard to fix this alone, but now help would be very appreciated.
I got it already running, but had only graphics with one homebrew demo. Some tiles are missing and it was really hard to find the reason for that. Now that my fix still doesn't work, I really need help.
All accesses to memory with _u16 and _u32 pointers are converted to return MSB value format and furthermore proper alignment is established.
Now my last problem - I thought - has something to do with incorrect register access.
Remember these bunch of mapB mapW and mapL files in the TLCS900 core?
Looking into these I saw accesses to 32bit register values with word and byte pointers, expecting to have it in LSB format. Of course the LSB access to the first byte of a 4 byte MSB register will return the last byte instead.
So I decided to modify these .h files to be compilable for both, MSB and LSB.
Now please, can you give some hints to me, what fixes you established regarding alignment and endianes?
Thanks,
Rockin'-B[/quote]
I suggest you look at the ngage port of neopop. if that doesn't help send me your email and i'll send you my code. But dreamcast is LSB so I dont know if my code would help or not.
Troy
My problem is fixed now, NeoPop runs on real Saturn, although the colors are slightly shifted to red.
As for the N-Pop source: I don't know what to think about it: Subbie made some changes to the windows version which one would only apply to a MSB port. But to run on an MSB machine, there are a lot more fixes to do.
Beware: subbie made some little experimental changes to the TLCS900 core, which do not behave like the original. I took only a part of them over, untill NeoPop runs correctly.
[quote author=RockinB link=board=dcemu;num=1098286979;start=60#67 date=01/06/05 at 13:23:39]
Beware: subbie made some little experimental changes to the TLCS900 core, which do not behave like the original. I took only a part of them over, untill NeoPop runs correctly.[/quote]
Yeah I didn't use those TLCS900 core changes, I only used the mem.c changes for 4 byte allignment and the gfx_scanline_colour, BGR swap to RGB .
Troy
[quote author=RockinB link=board=dcemu;num=1098286979;start=60#67 date=01/06/05 at 13:23:39]My problem is fixed now, NeoPop runs on real Saturn, although the colors are slightly shifted to red.[/quote]
Hmm, funny that I would read this on dcemu uk before I do on SX. I should probably check out the SX saturn boards, shouldn't I? ;)