emukidid
72cca5ba9a
Fixed the "rb" error when opening a file to write using libfat.
2009-10-14 12:34:24 +00:00
tehpola
1e4a087930
Added latest dynarec work to googlecode
2009-10-14 12:30:55 +00:00
tehpola
04a9c5ba92
Minor typo: I switched byte and half for the rw_nothing LUT; shouldn't
...
have affected anything.
2009-10-14 12:26:55 +00:00
tehpola
231e423702
Overhauled memory so that there's only one large memory LUT and a few
...
more very small LUTs.
2009-10-14 12:26:07 +00:00
tehpola
56df7d3be5
Implemented a correct (but not extremely accurate) sqrt.
2009-10-14 12:16:32 +00:00
tehpola
f7ea543f93
Commented some more DEBUG_print's which used DBG_USBGECKO in certain areas for a speedup.
2009-10-14 12:16:11 +00:00
tehpola
0604a015dd
Reenabled some previously disabled skip_jump code.
2009-10-14 12:13:06 +00:00
tehpola
db523931e6
Fixed GEN_FSEL macro.
2009-10-14 12:12:40 +00:00
tehpola
d28b9db21d
Reworked SQRT_FP to use less (and faster) instructions, and properly
...
handle sqrt(0).
2009-10-14 12:11:00 +00:00
emukidid
f02799e568
Savestates fixes ported from tehpola branch :)
2009-10-14 12:09:19 +00:00
emukidid
932ec0f8ab
Savestates fixed up - Working in dynarec and hopefully faster to save since I eliminated all the gzseek's :)
2009-10-14 12:08:19 +00:00
sepp256
4bdd439364
Reworked FB tex eviction to evict normal textures over FB textures that have been used in the past frame.
...
Reduced tex cache size for GC to 1.5MB again.
2009-10-14 03:51:29 +00:00
tehpola
848639323a
Don't try to recompile stores yet, it doesn't work.
2009-10-14 03:34:29 +00:00
tehpola
6752f8f0a0
Fixed genJumpPad bug which would cause delay slots to be executed twice
...
if the branch was the last instruction in a block and wasn't taken.
2009-10-14 03:32:50 +00:00
sepp256
22106fd171
Fixed compiler conditional in FrameBuffer.cpp
...
Upped GC tex cache to 2.5MB
2009-10-14 02:41:34 +00:00
sepp256
49a34cfdd5
Fixed FB tex crashes.
...
No longer evict current FB tex.
Increased Tex cache to 1.5 MB for GC so that Banjo Kazooie FB effect works.
2009-10-14 02:40:54 +00:00
sepp256
24dc8a5a96
Fixed menu gfx after running Earthworm Jim or Rampage 2.
...
General code cleanup.
2009-10-14 02:40:17 +00:00
sepp256
f4c437d3e4
Fixed flashing VI/s & FPS text for Zelda logo screen.
2009-10-14 02:39:35 +00:00
sepp256
e20cec10bd
Fixed flipped FB textures.
2009-10-14 02:38:57 +00:00
sepp256
313f818f83
More tweaking to EFB clearing.
...
Fixed crash from FrameBuffer Textures.
(Not sure if this fixes OOT menu crash for GC, yet.)
Fixed ROM cache loading bar.
2009-10-14 02:38:30 +00:00
sepp256
fd7a2702c1
Fixed timers. :)
2009-10-14 02:37:58 +00:00
sepp256
74791a7ecf
Changed EFB clearing to not require a ~200k buffer.
...
Now forcing EFB clear after every Dlist.
Copying EFB only when it has been drawn to.
Tehpola's branch untested, but should be fine.
2009-10-14 02:37:38 +00:00
tehpola
ed05e68a5e
Interpret jumps/branches whose delay slot lies in the next virtual
...
block
2009-10-13 21:57:45 +00:00
tehpola
d778f01e79
Fix FP_need_check stuff: always checks in delay slots to make sure the
...
old and in-place delay slots have the same size.
2009-10-13 21:47:04 +00:00
tehpola
e480ccf08a
dyna_check_cop1_unusable needs to address of the current instruction,
...
not the next one. I now pass the appropriate address when checking FP
unavailble.
2009-10-13 21:41:15 +00:00
tehpola
a0cef48c7d
Looks like I was making checking for code overlapping more complicated
...
than it needed to be and possibly not correct.
2009-10-13 21:39:45 +00:00
tehpola
b83bb65f67
Don't call nop_ignored when delay slots aren't compiled because they're
...
not branched to. Thus the delay slot should be NULL in code_addr which
will cause it to be recompiled from the delay slot on after wiping out
the old code.
2009-10-13 21:38:45 +00:00
tehpola
4b405cc7c2
Changed DMFC1 to behave like MFC1 which actually works (untested).
2009-10-13 21:38:09 +00:00
sepp256
b3cd1e57cf
Framebuffer Tex fixes.
...
EFB clearing rewrite.
Scissoring fix.
Tehpola's branch untested.
2009-10-13 21:35:30 +00:00
sepp256
b9095235a8
Revamped EFB clearing.
...
Reworked FB Textures.
Added FB Texture menu item to dev features menu.
Note: build in tehpola's branch not tested.
Note 2: FB Textures are sometimes unstable. I'm not sure why, yet.
2009-10-13 21:35:02 +00:00
tehpola
d9c5d46ab1
Fixed incorrect word-ordering in dmtc1.
2009-10-13 21:31:25 +00:00
emukidid
8eb0e3eb8f
Fixed saving to SD at least.
2009-10-13 21:22:09 +00:00
sepp256
81cf8f4712
Reverting the extra include in the Makefiles.
...
Just move #ifdef <asm.h> to the #ifdef LIBOGC_INTERNAL at the bottom of lwp_heap.h
2009-10-13 15:32:40 +00:00
sepp256
b54acef76a
Texture Cache now uses __lwp_heap to manage cached textures.
...
MEM2 is now used on Wii, so there's an extra 1MB of MEM1 free.
2009-10-13 15:29:01 +00:00
tehpola
d1d93d5444
RecompCache now uses its own __lwp_heap to manage its memory.
2009-10-13 13:13:52 +00:00
tehpola
92b61bbbd9
Recompile delay slots in place only if necessary.
2009-10-13 13:13:27 +00:00
tehpola
1ccaa38091
If a malloc or realloc fails, free up memory and retry.
2009-10-13 13:12:45 +00:00
tehpola
7d2d4ba5d9
Fixed recompiled load implementation, but there's some obscure bug left
...
in the stores.
2009-10-13 13:11:41 +00:00
emukidid
88fc5612fc
DVD Fixed up on GC (even for Qoob people like me).
2009-10-13 12:30:38 +00:00
tehpola
8dd6b14863
Initial attempt at recompiling all loads/stores.
2009-10-13 12:27:10 +00:00
tehpola
10795e74d6
In retrospect, there's nothing wrong with a shift amount of 0.
2009-10-13 12:26:49 +00:00
tehpola
bace968b44
Fix SLL, SRL, and SRA when SA is 0 and rd != rt.
2009-10-13 12:26:32 +00:00
tehpola
00947f0b47
Makefile.GLN64_dev fixed to compile with latest Dynarec :)
2009-10-13 12:26:11 +00:00
tehpola
59bbf221b7
Damn you, meld.
2009-10-13 12:24:56 +00:00
tehpola
863054de0c
Oops.
2009-10-13 12:23:22 +00:00
tehpola
23255c234d
I forgot to add/delete files.
2009-10-13 12:22:31 +00:00
tehpola
af7b8b93a2
Hopefully porting the entire dynarec to the public trunk code.
2009-10-13 12:21:43 +00:00
tehpola
e336380604
Fixed GC makefile Makefile.GLN64_dev to compile with latest code.
2009-10-13 05:24:34 +00:00
tehpola
21a8ed363c
Made logical instructions operate on 64-bit values.
2009-10-13 05:22:40 +00:00
tehpola
dcc9ace586
Allowing r11 and r12 to be mapped.
2009-10-13 05:21:24 +00:00
