Emulator

I am also having graphical glitches recently. I'm not sure but I think they could
have something to do with rodolfos viewport changes. That could also mean, though,
that only nvidia users have the issue...can you confirm that?

yup same here basically.

using a geforce 8800gtx and noticed the bloom effects are way off (the lights are
getting drawn above the source) also theres some tunneling effect where the
periferals look a bit like you are looking through a glass jar.

update i could get past the part where it crashes if i played in windowed mode with
the panic handler on and just keept spamming continue. well didnt help me much
though as i cannot change weapons and soon after you neeed to to keep out some
rather pesky necro unitologists before they rip you a new one xD

tried compiling todays svn first time it ran past the section that crashed second
time i got this.

Unhandled Exception
Code: 0xC0000005
Call stack info:
0x8001A92D : ?
0x068E1C68 : ?

Unhandled Exception
Code: 0xC0000005
Call stack info:
0x0723A92D : ?

now i cant even get past that door in windowed mode cause this crashes the emulator
completly ouch.

unfortunatly im not so good with C++ (guess i could learn it) mostly i code in C.

ok seems i fixed the bugger for now doing an extra check for valid memory

in memmap.cpp

in GetPointer

// GetPointer must always return an address in the bottom 32 bits of address space,
so that 64-bit
// programs don't have problems directly addressing any part of memory.
u8 *GetPointer(const u32 _Address)
{
// do we indeed have a valid memory pointer ?
if (!ValidMemory(_Address))
{
return 0;
}

and in IsRAMAddress

bool IsRAMAddress(const u32 addr, bool allow_locked_cache)
{
// do we indeed have a valid memory pointer ?
if (!ValidMemory(addr))
{
return false;
}

this seems to have fixed unknown memory pointer prefix errors for good i completed
dead space extraction yesterday :)

made a few other modifications as well in the wiimote plugin code theres a few places
with arrays with constant size 0 (big nono) but i guess it was done because the
correct values where not known instead of array[0] its better to do array[] or maybe
change them to pointers allbeit would have to use malloc then but then again that
might not be a bad idea.

if anyone want to try out the build i compiled both the x64 and win32 version with
the intel compiler.

ftp://90.184.233.166:21/dolphin.rar

ftp://90.184.233.166:21/dolphin_src.7z

also seems to have had an impact on the offset bloom effects ?

Please respond if this issue is still valid, or it will be closed.

What the heck ? Does this patch work ? Did it get merged ?