Search results

Tried guy fawkes libs with dev-c++. cant get it to work.

The first thing that breaks is the earphone socket(1st day). I had to solder it back. The next was the stick fell off (after a few months)I had to superglue it back. At the moment there is something rattling in my gp2x. I guess the stick and the nand will be next.

I'll have a look at it again. must be the new asm core. What was the last version when the other games worked? unfortunately it works fine with the only rom i have so its difficult for me to test. Can you do me a favour, I'll make a few different versions with different opcodes moved back to C...

what's wrong? It's quite likely to be due to the new asm core. I can have a look if I get some feedback.

Thats right. I'll try it

version q has about all the opcodes in asm. Now you can see a big difference in speed as you increase the fps. The video part of the emulation takes about 40ms per frame (17 per frame means full speed). So it still needs optimization in the graphics code.

if you do rebuild all, you need to change the makefile again as dev-c++ will redo the makefile

I've uploaded version p please try it out. It feels faster to me than the n version. I've been changing the c-core in arm-new.h to asm according to Normmatt's instructions / examples . Its about half done now. I believe that it can be optimized further if someone with better knowledge of asm...

manually change makefile.gp2x created by dev-c++ to be compiled by CC rather than CPP. It should complete. If you want I can zip and send to you by e-mail.

`SDL_GP2X_Display' was not declared in this scope `SDL_GP2X_MiniDisplay' was not declared in this scope these are from Paeryn's libs. should comment them out if you dont use PAeryn's libs problem with memgzio.c

I am trying it out now. I made a new project, closed the main.cpp which dev-c++ made automatically. copied all the source code files have to be careful with the parameter e.g. -DC_CORE -DSDL etc

Any suggestions how to improve this? Would it be better if the flags could be stored and restored as one 32bit value rather than separately? Is it better to set the flags in C or to change to assembler? #define OP_AND asm("and %0, %1, %2\n": "=r" (reg[dest].I):"r" (reg[(opcode>>16)&15].I), "r"...

I converting the cpu core to arm asm with Normmatt's help realeasing so people can see if there is any change. I guess I still need to work on this a bit more.

Actually I am not doing anything with gb,gbc part at the moment but its just a side effect from visualboyadvance so its good to have if it still works. It would make the code a little smaller if I cut it out but I am not sure if its worth the effort.

the one by guy fawkes works for me. SDL libs v211006

and these #define LOGICAL_LSL_REG \ {\ u32 v = reg[opcode & 0x0f].I;\ C_OUT = (v >> (32 - shift)) & 1 ? true : false;\ value = v << shift;\ } #define ARITHMETIC_RRX_REG \ {\ u32 v = reg[opcode & 0x0f].I;\ shift = (int)C_FLAG;\ value = ((v >> 1) |\ (shift << 31));\...

So can things like this be further optimized? /*#define OP_EORS \ reg[dest].I = reg[(opcode>>16)&15].I ^ value;\ \ N_FLAG = (reg[dest].I & 0x80000000) ? true : false;\ Z_FLAG = (reg[dest].I) ? false : true;\ C_FLAG = C_OUT; */ #define OP_EORS \ asm("eor %0, %1, %2\n": "=r"...

I've just uploaded (o) for the binary. It has a few more lines of asm so hopefully the difference will be more although I get the feeling that the original opcodes which Normmatt did were probably the most important. I would appreciate if anyone else who knows asm can give advice on...