Changes
"""""""
 The following describes the layout of this changelog.

 In releases with contributions from multiple developers:
---
 Version number
  Area of project changed: Description of change [developer]
---
 In releases with contributions from a single developer:
---
 Version number [developer]
  Area of project changed: Description of change
---

 Changes from older releases may not classify changes by area of project.


 Developer abbreviations:
  TRAC          Charles Bilyue'
  DBJH          Daniel Horchner


 0.853 [TRAC]
  PPU: Fixed handling of tiles clipped on both sides, thanks to
   Deathlike for the report
 0.852 [TRAC]
  PPU: Fixed scrolling in modes 5 and 6 when multiple horizontal
   pages were in use; thanks to powerspike and Deathlike for the
   report
  PPU: Fixed a bug which could cause improper behavior when dual-window
   logic was in use; thanks to Deathlike for the report which led me
   to this
  PPU: Fixed an issue which could cause writes to the $212B register
   to not take immediate effect
 0.851 [TRAC]
  Config: Fixed a bug where the 768x717 resolution setting was not
   restored properly from the config file
  PPU: Fixed a bug which caused OBJ tile upper-page address offset
   to be incorrectly added only for lower-page tile references
  PPU: Fixed a bug in screen addressing in offset-change mode
   that affected tiles with V-offset-change disabled
  PPU: Fixed option to disable offset-change emulation
  Screenshots: Fixed a bug where all screenshots saved were blank
  Source: Removed some debugging code that was no longer needed and
   could cause issues when the period key was pressed
  PPU: Fixed a bug which caused incorrect behavior when OBJ tile count
   exceeded their time limit

  Thanks go out to Deathlike and grinvader for their testing and
   numerous bug reports which helped lead to fixes in both the 0.85
   and 0.851 releases!
 0.85 [TRAC]
  5A22: Improved self-generated IRQ behavior, thanks to byuu for
   information
  PPU: Added some missing precision limitations to mode 7, thanks to
   anomie and byuu for research
  5A22: Fixed a bug where HDMA was being terminated one scanline too
   early
  Source: Fixed a few minor source issues that could lead to compiler
   warnings or errors
  PPU: Translated much of the graphics rendering code to C
  PPU: Fixed bugs in rendering of offset-change BGs when mosaic was
   enabled
  PPU: Added window clipping to OBJ and to EXTBG mode 7
  PPU: Implemented direct color mode for 8-bit pixels on BG1
  PPU: Implemented color window emulation, including fixed color and 
   sub screen arithmetic modes and most, if not all, of the various
   edge cases, thanks to Blargg for fast 15-bit blending algorithms
  ROM loader: Fixed a bug where IPS auto-patching would only work
   on the first ROM loaded
  PPU: Fixed mosaic support
  ROM loader: Fixed a bug in ROM allocation resizing which could
   cause crashes
  ROM loader: Fixed a bug where IPS auto-patching could cause
   incorrect memory mapping on LoROM
  ROM loader: Fixed a bug where IPS auto-patching could potentially
   cause bad patches or crashes when extending the ROM size
  5A22: Improved H-DMA behavior accuracy in a couple ways, thanks to
   byuu for information
  Config: Added default ROM directory support (must be set manually in cfg
   file, read the config file section of the readme for info)
  Config: Added full-screen/window configuration option (controllable
   via GUI and config file but not commandline, at this time), and
   a few new resolution options
  GUI: Slight improvement to load ROM dialog
 0.842
  SPC700: Added support for BCD fixup opcodes DAA and DAS, thanks to
   anomie and Overload for information and resources for research and
   testing [TRAC]
  SPC700: Added support for opcodes BRK and RETI, thanks to anomie for
   information [TRAC]
  SPC700: Corrected DIV behavior, thanks to anomie for research [TRAC]
  PPU: Added support for separate internal H-offset and V-offset registers
   for matrix rendering [TRAC]
  SPC700: Fixed a flaw in DIV algorithm, thanks to Heretic for pointing
   this out [TRAC]
  ROM loader: Added IPS auto-patching support (must be in same path as ROM
   image, in ROM image ZIP file, or in save path) [Nach]
  ROM loader: Continuing clean-up of ROM loading code [Nach,TRAC]
  SPC700: Improved opcode trace logging code [TRAC]
  5A22: Added opcode trace logging code [TRAC]
  Sound: Added noise source emulation, thanks to Shay Green's
   Game_Music_Emu for information [TRAC]
  5A22: Improved DMA timing accuracy slightly [TRAC]
  Sound: Changed DSP register reads to ignore address bit 7; thanks to
   byuu for relaying research from anomie [TRAC]
  Source: Fixed a typo in the code which could appear and prevent
   compilation in a currently unused conditional case, thanks to byuu
   for pointing it out [TRAC]
  PPU: Added support for returning open bus values from the PPUs [TRAC]
 0.841 [TRAC]
  Sound: Fixed a flaw in the order of operations in sample volume
   multiplication, causing a slight improvement in sound accuracy
  SPC700: Fixed a bug in the MUL YA opcode which caused it to not update
   flags
  SPC700: Fixed a bug in the POP PSW which caused the direct page base
   address to not be updated
  Render: Corrected handling of writes to BG offset registers and mode 7
   matrix registers, thanks to anomie for research
  Sound: Improved timing accuracy of direct envelope height setting mode
  SPC700: Corrected POP PSW to allow changing of B flag, and PUSH PSW to
   reflect its current state
  SPC700: Corrected a bug in updating of H flag by SBC, and some bugs in
   updating of flags in ADDW and SUBW, and potentially in INCW and DECW;
   thanks to Overload for some of the related research
  Sound: Fixed a bug which caused attack rate 15 to be far too slow
  Sound: Fixed bugs in sustain and release which could cause ENVX to
   underflow
  Source: Fixed some issues in the source which caused problems building
   on Linux, thanks to Jonathan Gevaryahu for pointing this out
 0.84 [TRAC]
  Source: Fixed a flaw that caused problems during linking
  SPC700: Fixed a bug in the timing of timer 2
  SPC700: Fixed a bug in the timing of SPC700 opcode execution
  SPC700: Fixed a bug affecting all of the SPC700 timers, causing the
   delay needed for cases where the timer target is set previous to the
   current position to only be applied in all cases where it was not
   needed
  SPC700: Rewrote entire core, replacing it with a cycle-based core
   written in C; helps Tales of Phantasia and games by Enix
  SPC700: Corrected timer counter registers to not be reset on write,
   and corrected most write opcodes to read first, thanks to anomie for
   research confirming this theory
  Source: Fixed a bug in the makefile that caused 'make clean' to not
   remove the binary
 0.833
  SPC700: Fixed timing of MOVW da,YA opcode, thanks to anomie for research
   [TRAC]
  SPC700: Added a brief delay before code execution begins approximating
   the time taken by reset [TRAC]
  5A22: Added DMA setup delay of two slow bus-cycles [TRAC]
  Timing: Mostly corrected H-IRQ and H-counter timing for positions
   after the 'long' dots near the end of scanlines [TRAC]
  Timing: Adjusted hardware interrupt timing, thanks to some info from
   DMV27 [TRAC]
  Timing: Fixed latch behavior with respect to I/O port; improved latch
   reporting code [TRAC]
  5A22: Moved WAI delay to after hardware interrupt detection, thanks to
   anomie for verifying this [TRAC]
  SPC700: Improved flag behavior of ADDW, SUBW, and MUL opcodes, thanks
   to anomie for research [TRAC]
  SPC700: Fixed some issues in timers [TRAC]
  SPC700: Improved timer accuracy and behavior of register area, thanks
   to research from anomie [TRAC]
  Source: Removed support for double-clocked SPC700 execution rate [TRAC]
  JMA Support: Fixed some bugs in decompression, improved performance,
   reduced memory requirements, upgraded stream support, and added
   support for JMA v1 files [Nach]
  Sound: Corrected BRR fetches across end of address space to wrap and
   continue fetching and to not terminate channel, thanks to anomie for
   research [TRAC]
  Sound: Adjusted some aspects of DSP reset [TRAC]
  Sound: Fixed a bug in decoding of BRR blocks with invalid range values;
   thanks to anomie for pointing this out [TRAC]
  Source: Improved some aspects of makefiles and source layout [TRAC]
  5A22: Added emulation of the capability of HDMA to be used for reading
   the B-bus, thanks to anomie for alerting me to its presence [TRAC]
  SPC700: Removed 'dirty' opcode fetching [TRAC]
 0.832
  Documentation: Updated grossly insufficient copyright attributations
   [TRAC]
  5A22: Corrected decimal fixup in ADC/SBC instructions to not
   allow multiple carries/borrows [TRAC]
  Input: Prevented input code from reading sticks that were not
   providing digital input due to needing calibration [TRAC]
  Memory map: Corrected a scarcely used area of the LoROM map [TRAC]
  Render: Disabled drawing of sub-screen layers when screen arithmetic
   disabled [TRAC]
  5A22: Fixed a bug in timing code that could cause crashes [TRAC]
  5A22: Fixed a bug which broke H-IRQ timing [TRAC]
  Timing: Moved display start to H=1 and hblank start to H=274, per
   recent research [TRAC]
  APU: Fixed a serious bug in OUTX reading [TRAC]
  Source: Changed memory access emulation code to be more cache-
   friendly [TRAC]
  Source: Fixed format of include path specification argument for GCC
   3.4.x [TRAC]
  Sound: Hacked in some code to try to make the Win32 port prefer the
   DSound mixer over the Allegro mixer, alleviates serious audio
   latency issues [TRAC]
  Misc: Added support for reading .jma files [Nach]
  ROM loader: Minor clean-up in ROM loading code [Nach]
  Memory map: Fixed a few bugs that could cause bad addresses to be used,
   sometimes leading to faults [TRAC]
 0.831
  5A22: Changed emulation-mode interrupts to NOT zero data bank
   register - thanks to Overload for pointing out that the SNES CPU
   does not do this [TRAC]
  Source: Fixed a bug which prevented compilation on some setups,
   including BeOS [DBJH]
  Sound: Added 8-sample delay before key-on completion [TRAC]
  Timing: Fixed up some IRQ setup logic bugs introduced by the
   NMI logic improvements [TRAC]
  Render: Fixed a few bugs where some window settings were not
   affected by reset [TRAC]
 0.83 [TRAC]
  Sound: Fixed a bug in envelope height reading of channels that weren't
   playing; value to be read wasn't being updated, fixes Secret of
   Evermore
  Sound: Improved envelope timing when envelope settings changed for
   playing voices
  Sound: Altered audio code to base timing on samples instead of cycles
  Sound: Added countdown register emulation to envelope and noise
   timing
  5A22: Corrected emulation-mode interrupts to zero data bank register
  Documentation: Corrected Win32 port compilation requirements, other
   minor improvements
  5A22: Corrected emulation-mode behavior of many opcodes
  5A22: Added variable bus speed emulation, and improved overall timing
  5A22: Fixed some bugs in DMA
  5A22: Changed DMA to no longer be able to access internal CPU registers
  5A22: Added open bus emulation to A-bus
  Render: Fixed a serious bug which could have some pointers used by
   offset-change rendering not be initialized
  PPU: Corrected a behavior of the H/V latches; reading register $2137
   (SLHV) does NOT reset which bits will be read next
  Timing: Changed scanline length back to 341 dots after research on the
   hardware gave this number; still stumped about SGB and Shin Megami
   Tensei
  Timing: Improved NMI emulation; now supports delayed NMI,
   as well as re-acknowledgement of NMI
  Timing: Added consumption of clock cycles by general DMA
  Timing: Changed default SPC700 CPU clock to 1.024MHz
  Sound: Changed APU code to build with update-on-RAM-change by default
  ROM loader: Added support for loading ROMs up to 64Mbit (autodetection
   still needs work)
 0.81
  PPU: Greatly simplied VRAM read/write handling and readded VRAM read
   buffering, thanks to anomie for convincing me to give it another
   chance, and Overload for some information which fixed some
   related glitches [TRAC]
  Timing: Implemented partial HDMA timing [TRAC]
  Render: Improved emulation of sprite priority rotation, thanks to
   anomie for research [TRAC]
  5A22: Improved RDIO emulation [TRAC]
  Source: Moved emulation code for registers internal to the main CPU
   ($4xxx) to main CPU source files [TRAC]
  Screenshots: Fixed a bug where screenshots were not properly saved in
   some screen modes [TRAC]
  SPC700: Fixed a serious bug in the handling of some 16-bit reads,
   thanks to Brad Martin for alerting me to the problem [TRAC]
  Render: Added countdown register support to mosaic effect in
   modes 0-6 [TRAC]
  Render: Fixed a tileset lookup bug in mosaic effect in mode 4 on BG1 
   with 16x16 tiles enabled [TRAC]
  Render: Fixed a register corruption bug in mosaic effect in mode 4 on
   BG2 with 16x16 tiles enabled [TRAC]
  SRAM: Fixed a problem where SRAMs smaller than 64kbit would not be
   mirrored on load [TRAC]
  Render: Fixed an off-by-6 bug in mode 7 vertical flip [TRAC]
  Input: Fixed a few bugs related to multiple 'sticks' in
   joystick/joypad code, thanks to Harekiet for reporting [TRAC]
  Render: Added correct limits to some intermediate calculations
   in mode 7; values (BG1HOFS - M7X) and (BG1VOFS - M7Y) actually
   produce 11-bit signed results, with the MSB taken from bit 13, not
   bit 10; fixes Space Colony escape in Super Metroid; thanks to
   Qwertie's Mode 7 Test which aided immensely in research [TRAC]
  SPC700: Implemented correct RAM initialization pattern (fill each
   byte with bit 5 of its address); thanks to Overload for research,
   and MKendora for informing me of the findings [TRAC]
  DMA: Corrected behavior of HDMA mode 5, thanks to anomie for the
   information [TRAC]
  Render: Added support for the undocumented OBJ size settings,
   thanks to Charles MacDonald and anomie for research [TRAC]
  Timing: Changed scanline length to 335 dots and adjusted frame timers
   to compensate, to match the 61.17Hz NTSC SGB framerate; fixes demon
   fusion in Shin Megami Tensei [TRAC]
  5A22: Improved open bus behavior in several registers [TRAC]
  Source: Eliminated need for building with a patched NASM on Win32
   port, thanks to Nach for getting me to write the object patcher
   [TRAC]
  Source: Fixed object tree autogeneration [TRAC]
  Logo: Another completely new logo! [Siloh]
 0.80
  Misc: Fixed a flaw where divide result registers were not changed
   by division-by-zero [TRAC]
  Input: Fixed a flaw in JOYC1 which caused SMB in Mario All-Stars to
   not acknowledge the start button [TRAC]
  Source: Seperated the makefile for the multiple ports [TRAC]
  Documentation: Cleaned up and clarified the history, and moved it
   to another file [TRAC]
  Ports: Added Linux/Unix build support [DBJH]
  Source: Contributed to the removal of some non-portable code and
   more source clean-up [DBJH]
  ROM loader: Reduced some problems with loading ROMs in non-absolute
   paths [DBJH]
  Ports: Added BeOS build support [DBJH]
  Configuration: Improved the portability of code which handled
   configuration file regeneration [DBJH]
  Source: Fixed a few problems in variable declarations [DBJH]
  GUI: Made ".." come first in the file list [DBJH]
  GUI: Fixed some problems where text would be displayed outside a
   window [DBJH]
  Source: Reorganized the source tree [TRAC]
  65c816: Fixed a serious design flaw in ADC/SBC emulation which caused
   the overflow flag to be set to an unexpected state [TRAC]
  Sound: Removed some extraneous and problematic compile-time directives
   [TRAC]
  Sound: Added clipping and saturation to gaussian interpolation filter,
   thanks to Jonathan Gevaryahu (Lord_Nightmare) and Brad Martin for the
   fixes [TRAC]
  Timing: Moved display period to the start of scanlines [TRAC]
  DMA: Fixed a bug where the B-bus address could not be changed to a value
   equal to the DMA parameter for the same channel [TRAC]
  Timing: Corrected normal display to 224 lines [TRAC]
  Timing: Added support for latching 213C/213D counter registers via
   register 4201 [TRAC]
  Misc: Added support for reading .zip and .gz files [DBJH]
  Render: Fixed a bug where dual-window AND logic would result in
   an entire line being clipped if either window covered the
   whole display [TRAC]
  GUI: Cleaned up some usage of constants in the source [TRAC]
  GUI: Moved some input-related code out of the GUI source [TRAC]
  Input: Added support for using joysticks/joypads to emulate
   controllers; must be set up in controller configuration in GUI
   similar to keyboard input [TRAC]
  Configuration: Changed format of scancodes for input mappings to
   decimal [TRAC]
  Timing: Improved PPU counter latch accuracy [TRAC]
  Render: Restructured windowing code [TRAC]
  Render: Added preliminary window clipping to mode 7 background [TRAC]
  Render: Improved layering order in mode 7, helps Final Fantasy 6
   world map [TRAC]
  SRAM: Corrected LoROM SRAM address ranges, per a theory shared by
   myself and MKendora [TRAC]
  Memory map: Corrected ROM mirroring [TRAC]
  ROM loader: Reduced some problems with starting the emulator in a path
   other than the one where it is located [TRAC]
  Render: Fixed some mode 7-related problems in Y's III and Illusion of
   Gaia; it appears that the intermediate values (BG1HOFS - M7X) and
   (BG1VOFS - M7Y) produce 10-bit signed results; thanks to SNES'9X
   for help [TRAC]
 0.776 [TRAC]
  Render: Added support for dual window clipping with all logic
   modes of backgrounds in modes 5 and 6
  Render: Added full mosaic effect to modes 5 and 6
  Render: Added support for offset change in mode 6
  Render: Fixed an offset map addressing bug in all offset-change
   renderers
  SPC700: Improved CPU cycle count conversion accuracy in PAL emulation
  Configuration: Removed support for old (pre-0.60) format .cfg files
   to help simplify the code and solve a few bugs
 0.774 [TRAC]
  SRAM: Added support for mirrored SRAM at banks F0 and above in LoROM
   memory map, helps Big Sky Troopers and others, thanks to MKendora
   for the information
  Render: Fixed a screen map addressing bug in 16x16-tile background with
   mosaic renderer
  Render: Fixed a pixel count computation bug in 16x16-tile background with
   mosaic renderer
  Render: Fixed a screen map addressing bug in 16x16-tile background with
   offset-change and mosaic renderer
  Render: Added support for dual window clipping with all logic
   modes of 8x8-tile backgrounds with offset change and mosaic in
   modes 2 and 4
  Render: Added support for dual window clipping with all logic
   modes of 16x16-tile backgrounds with offset change and mosaic in
   modes 2 and 4
  Interface: Changed filename prefix for screenshots
  SPC700: Fixed a flaw in SPC700 timers that could cause fractional ticks
   to be lost on timer startup
  SPC700: Fixed a bug which caused the POP PSW opcode to not restore the
   negative flag properly, thanks to Brad Martin for help with this
 0.77 [TRAC]
  SPC700: Fixed the SPC700 timer bug that was supposed to be fixed in
   0.75; not certain how it didn't make it into the release
  Sound: Fixed a serious bug which caused the sound to never be updated
   on DSP register changes; allows the correct attack 15 timing to
   work with FF4, thanks to MKendora and Brad Martin for getting me
   to go back to it
  Sound: Changed behavior of key on and key off registers, should
   be more accurate
  Render: Added support for dual window clipping with all logic
   modes of 8x8-tile backgrounds with offset change and no mosaic in
   modes 2 and 4
  Render: Added support for dual window clipping with all logic
   modes of 16x16-tile backgrounds with offset change and no mosaic in
   modes 2 and 4
  65c816: Fixed a bug in which caused the timing position of an IRQ on
   a new scanline to be incorrectly computed; fixes a problem in
   Tetris Attack
 0.751 [TRAC]
  Render: Disabled some debugging code that caused mosaic to be forced
   on in background rendering, thanks to Andrs Tangir for pointing
   this out!
 0.75 [TRAC]
  65c816: Corrected address wrapping in direct and stack addressing
   modes, thanks to MKendora for pointing out these problems
  65c816: Altered register allocation
  SPC700: Altered register allocation
  SPC700: Fixed a bug in the timers which caused them to be reset
   to random values when turned on, thanks to Brad Martin for pointing
   this out
  Source: Removed some obsolete code and dependencies
  Source: Started cleaning up many of the C label references in the asm
   code
  Render: Added support for dual window clipping with XOR or XNOR logic
   modes of 8x8-tile backgrounds with no offset change or mosaic
  Render: Fixed a bug where window clipping was not updated correctly
   when changing layering order
  Render: Added support for dual window clipping with all logic
   modes of 16x16-tile backgrounds with no offset change or mosaic in
   modes 0-4
  Render: Added support for dual window clipping with all logic
   modes of 8x8-tile backgrounds with mosaic and no offset change
  Render: Added support for dual window clipping with all logic
   modes of 16x16-tile backgrounds with mosaic and no offset change in
   modes 0-4
 0.74 [TRAC]
  Render: Altered clip window setup code
  Sound: Corrected behavior of invalid range values in BRR header byte,
   thanks to Brad Martin for the information
  Sound: Corrected bug in ADSR/GAIN switching; when switching from GAIN
   to ADSR, envelope update time was not set based on current ADSR state
  Sound: Restructed BRR decoding
  Render: Changed the back area color addition hack to actually add to
   color 0 instead of replacing it, helps many games
  Render: Added support for dual window clipping with OR or AND logic
   modes of 8x8-tile backgrounds with no offset change or mosaic
  Sound: Corrected sample loop behavior, erroneously changed it in 0.73,
   fixes Sidmania
  Sound: Added emulation of 4-point pitch-regulated gaussian
   interpolation of sample data
  Sound: Added emulation of SNES echo effect and its corresponding FIR
   filter
 0.735 [TRAC]
  PPU: Fixed some register corruption issues in VRAM write handlers
  PPU: Altered VRAM writes; when VRAM writes are attempted during the
   display period, update addresses but do not save data, fixes Mega
   Man X status screen glitch
  Sound: Modified the envelope update values
  Sound: Altered timing of attack rate 15, thanks to Brad Martin for
   the correction
  Sound: Altered behavior of ADSR envelope decay with the maximum
   sustain level, thanks to Brad Martin for the correction
  Sound: Altered ADSR/GAIN switching; when switching from GAIN to
   ADSR, use the last ADSR state used by the channel since key on,
   fixes Mega Man X music looping bug, thanks to Brad Martin for help
   with this
  Source: Fixed a bug in the makefile where some flags weren't being
   properly passed for building the CPU cores
 0.73 [TRAC]
  Timing: Changed Korea region to South Korea, and its video standard
   default to NTSC, fixes Dragon Ball Z - Super Butoden 3 (K), thanks
   to MKendora for the information!
  Sound: Changed envelope behavior: when voices already on are keyed on,
   reset volume to zero, thanks to Brad Martin for setting me straight
   on this
  Sound: Fixed envelope behavior: no longer allowing envelope mode
   changes for voices that are releasing
 0.725 [TRAC]
  Render: Added support for single window clipping of 8x8-tile
   backgrounds with no offset change or mosaic
  Timing: Fixed the Final Fantasy 6 timing hack
  Render: Fixed a minor bug in offset change screen map address
   calculation
  Render: Corrected address wrapping for wide and tall screen maps
   at the end of VRAM
 0.721 [TRAC]
  Render: Removed some unused masking code for performance
 0.72 [TRAC]
  Memory map: Added a hack for Batman: Revenge of the Joker, an odd
   ROM with LoROM memory map and ROM info block in the 2nd 32k bank
  Memory map: Fixed behavior of command-line options -fi (Force
   interleaved) and -fn (Force non-interleaved) on non-split ROMs when
   -fl or -fh also used
  Memory map: Fixed behavior of command-line options -fl (Force LoROM)
   and -fh (Force HiROM)
  Timing: Added a hack to eliminate the world map flicker on Final
   Fantasy 6 until timing is improved
  Source: Cleaned up some parts of the source and altered makefile to
   help portability
  Sound: Reevaluated and adjusted sample ranges to increase output
   volume
  GUI: Added options for independently stretching the emulation
   display output in the horizontal and vertical directions
  Other: Added blitter support for the stretch options in 16-bit
   screen modes
  Render: Added full mosaic effect to modes 2 and 4, 16x16-tile
   backgrounds with offset change
  Sound: Re-enabled pitch-modulation
  Sound: Changed the BRR sample decoding to a new method based on
   information from Brad Martin, helps sounds in several games
   (Final Fantasy 6, Chrono Trigger, Mega Man 7, Mega Man X, others)
  Render: Improved performance of background rendering in modes 5 and 6
  Configuration: Reordered screen modes due to new stretch options
 0.706 [TRAC]
  65c816: Added support for the inability to acknowledge an IRQ in the
   instruction immediately after one that clears the I flag, fixes
   Marko's Magic Football, thanks to anomie for information on the
   problem
 0.705 [TRAC]
  65c816: Optimized portions of the main execution loop
  SPC700: Optimized portions of the main execution loop, cycle
   conversions, and timer calculations
  Memory map: Removed the hack for Mega Man X/Rock Man X, it was
   mistakenly commented out some time ago, but is no longer needed
  Memory map: Added support for ROM mirroring, fixes
   Mega Man X/Rock Man X
  SRAM: Corrected addressing in 512kbit and 1Mbit SRAMs, fixes
   Dezaemon and Thoroughbred Breeder 3
 0.702 [TRAC]
  Other: Added code to use the FPU or MMX to clear the framebuffer,
   and to allow disabling of cache preloading
  Config: Added options control usage of FPU/MMX and cache preloading
  Source: Fixed a few minor issues which prevented compilation with
   NASM 0.98.34
  SRAM: Added support for SRAMs larger than 512kbit - unfortunately,
   something must be off as nothing requiring it seems to work yet
  SRAM: Fixed a serious bug in support for HiROM SRAM larger than 64kbit
  Render: Fixed a bug which caused 8x8-tile offset change layers without
   mosaic to not display correctly in mode 4
  Render: Fixed a bug where tile priority was unlikely to behave
   properly in 8x8-tile offset change layers with mosaic
 0.701 [TRAC]
  Sound: Fixed some issues in GAIN register emulation
  Sound: Fixed some issues related to envelope updates
  65c816: Fixed a bug in 16-bit ADC/SBC where invalid values could
   sometimes be stored in the accumulator, fixes a crash in Terranigma
 0.70 [TRAC]
  Sound: Fixed a serious bug in ENVX reading, fixes samples being cut
   short in Super Offroad: The Baja
  Sound: Fixed a serious bug in mixing epilog, fixes freezes in
   Terranigma
 0.694 [TRAC]
  Render: Fixed a bug which caused char 0 repeat outside the screen
   area to not work
  Render: Added full mosaic effect to modes 2 and 4, 8x8-tile backgrounds
   with offset change
 0.692 [TRAC]
  Render: Fixed a bug in modes 5-6 where screen map selection was
   sometimes incorrect when 64-tile screen width was used (fixes
   Seiken Densetsu 3 menu screens)
  Timing: Fixed a bug where IRQs could be set for non-existant H-timer
   positions
  Render: Added full mosaic effect to modes 0-4, 16x16-tile backgrounds
   without offset change
 0.69 [TRAC]
  SPC700: Eliminated some self-modifying code
  Timing: Fixed an IRQ-related bug in register $2137 (SLHV) (thanks to
   SNES'9X)
  Render: Added full mosaic effect to mode 7
  Render: Added full mosaic effect to modes 0-4, 8x8-tile backgrounds
   without offset change
  Timing: Changed timing of RDNMI/HVBJOY vblank start closer to NMI
   like it was before 0.68 - fixes ActRaiser title
  PPU: Eliminated some self-modifying code
  PPU: Fixed a bug in VRAM access (fixes some issues on Earthworm Jim 2
   copyright screen)
 0.68 [TRAC]
  Input: Fixed a bug in the input code
  GUI: Added the ability to configure shift keys for controllers
  Source: Fixed a makefile issue where non-profiling builds were
   linked with profiling enabled - it is safe to delete the
   gmon.out files that were being produced (thanks to stainless
   for reporting)
  Render: Overhauled 8x8 tile background plotters
  65c816: Fixed several address wrapping bugs
  65c816: Optimized 16-bit loads
  SPC700: Fixed a bug in POP PSW where the 'direct page' offset was
   sometimes incorrectly set, fixes Super Punch-Out!!
  SPC700: Various minor bug fixes and optimizations
  SPC700: Optimized 16-bit loads
 0.666 [TRAC]
  Render: Fixed a bug which was causing BG3 high priority tiles to not
   be drawn in mode 1 when BG3 highest priority was disabled
  Render: Restructured the 16x16 tile background plotters
  Render: Added support for offset change in modes 2 and 4,
   16x16-tile layers
  Render: Fixed a bug in the offset-change elimination speed hack
   (was causing it to sometimes disable offset-change despite it
   being used)
  Render: Corrected address wrapping for the vertical offset table in
   mode 2, fixes WWF Wrestlemania Arcade
  PPU: Altered background offset register behavior, fixes Theme Park
   (thanks to anomie and Matthew Kendora)
  Sound: Changed the behavior of the BRR decoding rate for noise-enabled
   channels
  Render: Fixed an OBJ setup bug for large OBJ partially off the left
   edge
  GUI: Fixed a bug where the logo wasn't always displayed
  Logo: Thanks to stainless for providing a new logo!
 0.665 [TRAC]
  Render: Corrected OBJ range limitations (thanks to anomie and Neviksti)
  Render: Removed BG3 highest priority from mode 0 since it doesn't
   actually exist (thanks to anomie)
  Render: Readded mode 7 rendering for the sub-screen
  SPC700/APU skipper: Added APUI00-APUI03 port mirroring
 0.662 [TRAC]
  Render: Corrected mode 7 EXTBG (priority-per-pixel) emulation
   (thanks to anomie and Neviksti)
  PPU: Corrected OAM register emulation (thanks to anomie and Neviksti)
 0.66 [TRAC]
  Sound: Fixed sound DSP key-on behavior
  Sound: Fixed a memory leak when sound was disabled
  Sound: Fixed crashes when switching sample sizes
  Sound: Fixed some issues with 8-bit sound
  Sound: Overhauled portions of the mixing code
  Render: Enabled some portions of the tile cache code which should have
   been already
  Source: Overhauled the makefile
  Source: Reorganized background rendering code
  Render: Added support for offset change in mode 4, 8x8-tile layers
  Render: Added code to eliminate emulation of offset-change when it
   is unused
 0.653 [TRAC]
  Sound: Fixed sound<>SPC timing speed, the conditional-compile directive
   was placed a few lines too late
 0.652 [TRAC]
  Sound: Fixed a sound DSP behavior issue - DSP address bit 7 is
   significant (was ignored since 0.625), fixes Donkey Kong Country
   series, others
 0.651 [TRAC]
  Render: Fixed a few cases where a segment register was not properly
   restored
  Render: Cache-related optimizations in mode 7
 0.65 [TRAC]
  65c816: Optimizations in flag handling and memory access
  SPC700: Optimizations in memory access
  Render: Altered method of decoding 4- and 8-bitplane tiles
  Render: Altered method of clearing framebuffer
 0.641 [TRAC]
  Source: Removed some debugging code that was no longer needed
  Source: Removed errors and warnings when compiling under GCC 3
 0.64 [TRAC]
  Render: Simplified/reoptimized mode 7 rendering
  Render: Corrected depth of low priority sprites in mode 7
  Timing: Fixed a bug which caused some events to occur one instruction
   too late
  Render: Overhauled sprite limitations and priority rotation,
   fixes Final Fantasy: Mystic Quest, perhaps others
  SPC700: Optimizations in flag handling
 0.63 [TRAC]
  Render: Reimplemented priority-per-pixel in mode 7
  Render: Fixed a bug that caused some calculations to be done too
   often in mode 7
 0.625 [TRAC]
  Sound: Removed some debugging code that was breaking Secret of
   Evermore
  Sound: Altered some of the envelope update timings
  Render: Fixed a mode 7 off-by-one-line bug
 0.62 [TRAC]
  Sound: Fixed severe slowdown when sustain was set to infinite
  Sound: Fixed severe slowdown with noise
  65c816: Fixed a crash bug related to WAI instruction
  Render: Fixed a bug where sprites were not recomputed when highest
   priority sprite was changed (fixes star sprite test in test ROM)
  Timing: Changed first non-vblank scanline to 0
 0.61 [TRAC]
  Frameskip: Fixed two bugs in the speed-throttle code: one when speed
   limiter was disabled (min frameskip > 0), one when fast-forward was
   used
  Render: Added code to eliminate emulation of unneeded priorities
  Render: Reoptimized and overhauled the general structure of mode 2
   offset change emulation
  Render: Fixed a bug in mode 7 outside-screen-area clip mode
  GUI: Fixed the behavior of the FPS counter for FPS > 99
  GUI: Added a 'rendering break' counter - tells how many times the
   delayed renderer is forced to synchronize rendering in the last frame
  GUI: Modified the keyboard handling to solve stability issues caused
   by a problem in the DJGPP/Allegro code - GUI is again enabled by
   default
  Sound: Fixed the register behavior of disabled voices
  Sound: Fixed sample looping when loop position changed
  Sound: Improved performance of sound generation
  Sound: Added option to allow generation of 8-bit output samples,
   and a GUI option to toggle sample size
  Sound: Added a config file option to select the Allegro mixer mode
  Configuration: Added 320x240x256 VESA2 screen mode as mode 8
 0.60 [TRAC]
  Render: Fixed a bug in mode 7 vertical-flip
  Render: Rewrote sprite rendering again - improves speed
  Configuration: Restructured config file
  Render: Added independent horizontal/vertical stretch options
  Source: Now linking with Allegro 3.9.34
  DMA: No longer preventing HDMA register from being written to during
   display - fixes some graphics glitches in Chrono Trigger
  PPU: No longer allowing writes to VRAM and OAM outside v-blank,
   force-blank
  PPU: No longer allowing writes to CGRAM outside v-blank, h-blank,
   force-blank
  65c816: Modified cycle counter and wait-for-interrupt emulation
  Timing: Improved 65c816<>SPC700 timing slightly
  GUI: GUI is now by default DISABLED because of instability - those who
   still wish to bear with it can enable it via a commandline switch
  Commandline: Added several new configuration switches
 0.55 [TRAC]
  Sound: Many changes to volume envelope timing
  Sound: Corrected synchronization of sound mixing and volume envelope
   generation
  Sound: Implemented noise generation
  Sound: Changed sound to use a double-buffering method - sound may
   skip on slower systems
 0.53 [TRAC]
  65c816: Fixed a bug in the MVP/MVN opcodes that could seriously
   disrupt timing (fixes Final Fantasy: Mystic Quest)
 0.52 [TRAC]
  Stability: Fixed a serious bug in the tile decoder that has caused
   crashes since the decoder was rewritten in 0.40
  Timing: Corrected an error in the CPU timing
  Interface: Added a fast-forward key (tilde)
 0.51 [TRAC]
  Render: Fixed a bug that caused some sprites to be corrupted
 0.50 [TRAC]
  65c816: Completed the memory mapper changes started in 0.49
  Source: Moved some debugging-related code into its own file
  Source: More source cleanup
  Stability: Fixed bugs in SRAM loading/saving (files opened weren't
   being closed) - appears to have fixed the last of the exit crashes
  Config: Changed the default save RAM file extension to SRM, and
   removed support for the old 'frameskip' setting in the config
   file
 0.49 [TRAC]
  Render: Implemented BG3 highest priority in mode 0
  Render: Corrected wrapping from end of tilesets
  GUI: Now uses standard keyboard controller auto-repeat
  Interface: Emulation no longer pauses when pressing certain keys
   (ESC to go to GUI, F1-F8 to toggle sound channels, 1-8 to change
   render options, 0 for screenshot)
  65c816: More changes to the memory mapper, for speed, accuracy, and
   allowing execution of code from SRAM (does anything use this?)
 0.48 [TRAC]
  Source: Ported the sprite decoder/plotters to NASM
  Source: Ported the background layer decoder/plotters to NASM
  Source: Ported the hardware port handlers to NASM
  Render: Removed a sprite 'fix' I put in back in 0.47 - did nothing
   except slow things down
  Display: Removed an unnecessary part of the screen mode setup code
  Render: Fixed a bug that caused sprites to be disabled when mode 7
   layer was disabled
  Memory map: Added support for the special memory map used by
   Rock Man X and Mega Man X (allows uncracked ROMs to work properly)
 0.472 [TRAC]
  Source: Ported the tile decoder/plotters to NASM
  Startup: Pulled out some debug code that caused the emulator to
   silently wait for a keypress at startup
 0.47 [TRAC]
  GUI: Extended the width of the Load ROM window (per request)
  Source: Ported the mode 7 emulation to NASM
  Display: Tweaked with the 256x239x256 setup code - appears to have
   been causing problems under some conditions
  Config: if it can be detected that a full path was supplied to start
   the emulator, it will now force the current directory to that path
   (temporary fix for Windows Explorer drag-and-drop startup until I
   do it a more elegant way) - ensures SNEeSe can find its .cfg and .dat
   files
  Config: Fixed some of the defaults used when a config file isn't
   present
 0.46 [TRAC]
  GUI: Fixed a few bugs that didn't allow the mouse to be enabled/used
  Source: Ported the DMA emulation to NASM
  Timing: Fixed a source inconsistency bug which could cause
   unpredictable SPC timing
  CPU: Altered the main memory mapper, should be faster than before
  Source: Ported the SPC700 CPU core to NASM
  65c816: Went back to more accurate stack operations in native mode
  Source: More cleaning up under the hood
  Render: Now clearing lines between actual end of display and maximum
   end of display
  Stability: Changed the way the internal render surface is transferred
   to the display - appears to have fixed the crashes during exiting
   that have been around since 0.43
 0.45 [TRAC]
  Sprite: Fixed sprite intermediate render buffer clearing -
   fixes garbage sprites in Bill Laimbeer's Combat Basketball, probably
   others
  Source: Fixed source to compile with and now compiling with
   GCC 2.95.2, Allegro 3.9.32 WIP
  NOTE: All keymaps set in cfg files from previous versions are
   incompatible, and must be reset!
  Stability: No longer using GCC asm labels, as GCC doesn't seem to be
   handling them correctly causing stability problems
  Stability: Wasn't always setting es register up for string instructions,
   fixed
  DMA: Now updating DMA transfer size after PPU->CPU DMA
  Timing: Made several changes to improve IRQ function - fixes SF2
   Turbo, New Ninja Warriors, probably others
  Source: Ported the 65c816 CPU core to NASM
  Stability: Noticed I wasn't locking a var used in interrupt context,
   fixed
  Allegro: Disabled Ctrl-Alt-End/Ctrl-Break exiting until I can figure
   out why they sometimes restart my computer under Win95
  Timing: Changed the way CPU cycle counts are translated into SPC
   cycles - should allow for closer CPU<>SPC synchronization
 0.441 [TRAC]
  APU skipper: Fixed a nasty bug that caused the emulator to crash
   upon entering the GUI if the APU skipper was enabled
 0.44 [TRAC]
  Render: Implemented correct sprite priorities (helps MANY ROMs:
   Chrono Trigger, Super Mario Allstars, Terranigma to name a few)
 0.43
  Logo: Thanks to neptron (http://ciot.pair.com/neptron/) for providing
   a new logo! [neptron]
  Disclaimer: Updated the disclaimer with more information on how
   SNEeSe may and may not be distributed [TRAC]
  Render: Wasn't resetting deinterleaved VRAM during Reset, now doing so
   (combined with resetting VRAM to 0, fixes missing tile on ActRaiser
   mode7 logo) [TRAC]
  PPU: Implemented delay-by-one-word read, from SNES'9X
   (was hoping it would fix Captain America... no luck) [TRAC]
  ROM loader: Added support for split ROMs (*.1) [TRAC]
  GUI: Fixed the controller configuration window (joypad pic was missing)
   [TRAC]
  Render: Fixed an offset change map address computation bug;
   helps vertical offset change in mode 2 (Tetris Attack looks much better)
   [TRAC]
  Render: Added tile priorities to offset change enabled mode 2
   (helps Art of Fighting) [TRAC]
  Render: Implemented priority per pixel in mode 7 (helps F1 Grand Prix)
   [TRAC]
  Render: Fixed a possible offset change map address wrap bug [TRAC]
  Render: Added horizontal scrolling support to mode 2 offset change
   renderer [TRAC]
  Render: Added screen size support to offset change map in mode 2
   offset change renderer (fixes Chrono Trigger menus!) [TRAC]
  Render: Added horizontal offset change to mode 2 offset change renderer
   [TRAC]
  Sound: Implemented custom mixer - improves sound greatly in most cases
   [Brad Martin]
  HDMA: Implemented HDMA-start-within-frame (fixes Killer Instinct)
   [TRAC]
  Render: Changed the way the internal bitmap is stored/handled
   (helps mode-X speed, allows screenshots in linear VGA modes) [TRAC]
  GUI: Worked on the Load ROM interface a bit - '.' dir is now excluded,
   directories are updated in their entirety (not one window at a time),
   sorting is done (both in names, and in types - files are at the top,
   followed by dirs, then drives) you can use the HOME and END keys now,
   and drive switching can be done (A:-Z: are listed at the far bottom)
   -note: for directories larger than 998 entries, only the first 998
    entries will be read! [TRAC]
  Sound: Now allowing ENVX (envelope height) reading to be disabled in
   the GUI (helps Bomberman series) [TRAC]
  Sound: Fixed a bug that allowed the sound code to be used when the
   SPC was disabled [TRAC]
  Sound: Hacked in a fix to prevent some ROMs (Bill Laimbeer's Combat
   Basketball) from crashing the emulator [TRAC]
 0.41
  Render: Added 3-way layering order cycle, press 8 key [TRAC]
  Timing: Made a few tweaks that may improve IRQ timing [TRAC]
  Sound: Implemented cycle-exact ADSR calculation, this fixes or improves
   sound on some things (Rampart), worsens on others [Brad Martin]
 0.40 [TRAC]
  65c816: Fixed a bug that could cause WAI to miss a waiting IRQ
  Source: Many changes to reduce repetition in 65c816/SPC opcodes
  65c816: Made a few improvements to the register allocation
  Sound: Fixed a possible buffer underrun bug
  PPU: Went back to the old VRAM read method - 'VRAM read buffering'
   might make more sense, but it doesn't work well with ActRaiser...
  PPU: Fixed a bug in VRAM full address increment (untested...)
  Render: Completely changed the tile caching method
  DMA: Made a few slight changes to the DMA/HDMA emulation
  Timing: Rewrote event handling, now using more correct timing for
   rendering and HDMA
  Timing: Implemented pre-NMI delay (fixes Chrono Trigger intro freeze)
  Timing: Fixed controller ready bit in HVBJOY (fixes Chrono Trigger
   combat slowdown and combat menu flickering)
    Thanks to Neill Corlett for info!
  Timing: Implemented much more correct H-IRQ timing, should also allow
   multiple H-IRQ's on a scanline
  Source: Ported some of the source from GNU-AS (GAS) to NASM
   (memory mapper, APU skipper, special screenmode setup)
  Joypad: No longer reporting parodoxical inputs - up and down can
   no longer be pressed at the same time, nor can left and right
   (fixes a potential problem in Zelda 3 name entry)
  SPC700: Now updating SPC timers at least once every Vblank
  SPC700: Fixed a major bug that caused the SPC to freeze after about
   17.5 minutes of virtual execution
 0.371 [TRAC]
  Config: Added save directory support (must be set manually in cfg file,
   read the config file section of the readme for info)
 0.37
  PPU: Modified the way VRAM read is done - now uses 'VRAM read
   buffering' which makes MUCH more sense, thanks to Ben Parnell
   of FCE Ultra (NES emulator) for info! [TRAC]
  Sound: Made a few improvements to the sample decompression routines
   [TRAC]
  Sound: Figured out why Soundscape cards weren't getting any sound
   and told Allegro to put the drivers in [TRAC]
  Sound: GUI should now be quiet on more cards [TRAC]
  CPU: Completely redid CPU opcode timing tables - thanks to Neill
   Corlett for better timing info! [TRAC]
  CPU: More improvements to the execution loop [TRAC]
  HDMA: Removed some unnecessary code that was in since the last rewrite
  Source: Removed an erroneous extra line in one file [TNSe]
  Source: Added a kludge in many files to work around the DJGPP linker's
   lack of object file alignment [TRAC]
  SPC700: Changed flag format conversion to use lookup tables [TRAC]
  65c816: Fixed a few bugs related to the Break flag in emulation mode
   (allowing Break flag to be cleared, Break flag set on stack after NMI)
   [TRAC]
  65c816: Simplified stack operations [TRAC]
  PPU: Fixed a bug that caused CGDATAREAD to be disabled [TRAC]
  Commandline: Added support for video standard forcing (-fvp to force PAL,
   -fvn to force NTSC) [TRAC]
 0.36 [TRAC]
  CPU: Timing base changed
  CPU: Redid the FastROM support, improving speed in cycle counting
  IRQ: Improved IRQ support (IRQs were being ignored in Vblank before,
   fixed)
  PPU: Improved joypad hardware emulation ('automatic read' timing)
  SPC: Improved speed in opcode fetcher, some source changes
  SPC: Now only resync'ing on CPU-SPC communication and on the start
   of Vblank (when sound update is done), doesn't appear to worsen
   anything and improves speed
  Other: Redid some of the debug dump from when the emulator exits
 0.35 [TRAC]
  SPC: Many more optimizations and some source cleanup
  SPC: Fixed yet another bug in the CPU-SPC cycle calculation code -
   caused Reset to cause 'unpredictable behavior' when the SPC was enabled
 0.34 [TRAC]
  Render: Slight tweaks inside the renderer, including a modification
   to the back area color addition hack
  Render: Implemented mosaic effect (vertical direction only)
  Render: Major tweaks to the vertical offset change support in mode 2 -
   still not correct but appears better
 0.33 [TRAC]
  SPC: Fixed a very stupid, nasty bug that showed up in 0.26, that
   caused the DIV YA,X opcode to corrupt the stack pointer
  Sound: Fixed a bug that caused the sound driver not to be installed
   if a ROM filename was not specified on the command line
 0.32
  Sound: Fixed some problems that caused sound generation to be used
   when sound was disabled [TRAC]
  SPC: Fixed a bug that caused SPC timers to lose cycles in conditional
   branches [Brad Martin]
  SPC: Now allowing SPC timer targets to be set when timer is enabled
   as it seems to fix other timing problems [Brad Martin]
  Render: Fixed some problems with layers using 16x16 tiles having
   small virtual screens [TRAC]
 0.30
  GUI: Fixed yet another small GUI bug [TRAC]
  Joypad: Tore out the old joypad code [TRAC]
  Allegro: Now linking with Allegro 3.11, should fix some sound card
   incompatibilities [TRAC]
  SPC: Bug fixes in 4 SPC opcodes, and a serious bug in the SPC CTRL
   register fixed, AND a new sound DSP/sound generation core
   (Compatibility noticably improved!) [Brad Martin]
 0.277 [TRAC]
  SPC: Fixed timer 2 to be a 64kHz timer - had mistakenly made it 8kHz
   in the rewrite
  SPC: Fixed some bugs in the sound DSP, that could have potentially
   caused memory overwrite
 0.276 [TRAC]
  APU skipper: fixed a nasty bug that showed up in (and went unnoticed
   since) 0.27, caused severe SNES timing problems
  GUI: Fixed a bug causing the CPU Reset to freeze up when the SPC is
   enabled after it having been disabled during execution
 0.275 [TRAC]
  SPC: Many more optimizations to opcode fetcher
  SPC: Implemented TCALL instruction
 0.27 [TRAC]
  SPC: More optimizations
  SPC: Completely rewrote SPC timers - fixes some SPC problems and sound
   playing too fast or slow
  SPC: Forcing SPC to 'catch up' once per 65816 'scanline' time period,
   may clear up more sound quirks (until sound generation code can be
   rewritten)
  Timing: Vblank timing adjusted slightly
 0.26 [TRAC]
  HDMA: Fixed a bug where the transfer size for the first line of a set
   was corrupted
  HDMA: No longer repeats write to register for continue mode
  IRQ: Put in a better H-IRQ hack, until I do the timing
  Render: Palette is set for highest brightness in the frame (helps
   Chrono Trigger, SF2 series, others), thanks to an anonymous
   contributor for help
  Timing: Hblank timing code simplified
 0.251 [TRAC]
  Render: Fixed the back color add hack (Mario World blue sky, others)
   that got broke somewhere along the way.
  Source: Fixed some problems in the makefile that gave some users
   trouble rebuilding.
 0.25 [TRAC]
  CLI: Implemented a basic commandline interface, providing a means to
   override ROM loader autodetection
  Released to the public!
 0.25b17 [TRAC]
  Timing: Reworked how Hblank is handled
  SPC: Now forced to 'catch up' once a frame, solves the 'lags' and
   improves sound overall
 0.25b16 [TRAC]
  SPC: Source cleanup, general optimization (register allocation needs
   more improvement), a few more fixes
  DMA: Fixed a bug where DASL/DASH was not being reset to 0 after general
   DMA transfer (thanks to ESNES and Trepalium for pointing me to DMA -
   F-Zero is fixed!!!)
  Render: Fixed a stupid, nasty, bug in the new sprite code
 0.25b15 [TRAC]
  GUI: Fixed another bug with changing SPC enable (menu item wasn't changed)
  Render: More changes to sprite render code
  PPU: Slight change to 16x8 multiply
  PPU: Fixed the emulation of the RDNMI register/timing - thanks to
   archeide of SNEmul for pointing me in the right direction!)
  65816: Fixed the write order of some instructions
  SPC: Now treating SLEEP and STOP as invalid opcodes for debugging purposes
   (is there any point in emulating these, similar to the 65816's STP?)
  SPC: Fixed part of the SPC register dump
  SPC: Implemented hidden 64 bytes of RAM at top of SPC address space
 0.25b14 [TRAC]
  Fixed a bug in the GUI with changing SPC/sound enable
  Added a FPS counter - it works, unfortunately you can't always see it...
   must be enabled in the GUI
 0.25b13 [TRAC]
  Reversed plot order back to the way it used to be before 0.16 - another
   speed increase (doh!)
 0.25b12 (unreleased) [TRAC]
  Implemented deinterleaved VRAM storage and a new mode 7 renderer:
   VRAM is now also stored as 32k low bytes and 32k high bytes. This has
   minimal effect overall but may provide a mode 7 speed increase, in
   particular for low-end systems (486)
  Fixed a bug where disabling mode 7 also disabled sprites
  More layering priority fixes in modes 2-6
  Altered the 256x240 mode to get rid of a lot of the excess border
  Reduced the 256x240 mode to 256x239
  Modified the way screen modes are handled with the GUI
  More tweaking with DMA/HDMA emulation
 0.25b11 [TRAC]
  Tweaked with the screen core a bit, removing a lot of redundant code
  Improved the header detection in the ROM loader (thanks to Gridle for the
   suggestion)
  Modified the way frameskipping is handled
 0.25b10 [TRAC]
  Tweaked with the controller code a bit - now DKC and Mario All Stars
   work as expected
 0.25b9 [TRAC]
  Fixed a bug in VRAM read ports - they were suggested by zsKnight as a
   likely cause of gfx corruption, and making a change to match a noticed
   difference in the '9X source, gfx corruption went away in the following
   ROMs, and likely others: ActRaiser, FF2, Super Metroid, Wario's Woods...
  Fixed a stupid bug that showed up in 0.25b8 that stopped FF2 from starting
  Several layering priority fixes
 0.25b8 [TRAC]
  Removed H-IRQ hack as it was causing more problems than it helped
  Implemented partial support for delayed IRQ
  Implemented a hack in WAI which causes the CPU timing to skip ahead to the
   next event
  Some SPC optimizations
  Another doc was wrong: increased SPC RAM back to 64k, sound is back to
   being reasonable (timing is still terrible)
 0.25b7 [TRAC]
  More CPU speed-ups
  Hooked SPC back up with on-demand execution, however sound is not
   considered a 'demand', and SPC is slow and buggy still.
 0.25b6 [TRAC]
  Layering priority fix: BG3 highest priority in mode 1 - BG3 high priority
   tiles now drawn over highest priority sprites, not under (helps Mario
   World map)
 0.25b5 [TRAC]
  Fixed a bug in full VRAM address increment
  Fixed a bug that sometimes caused Reset to freeze emulator
  Fixed mode 7 outside-screen-area clip/char 0 repeat - it appears that
   two docs were wrong on this...
  Native mode XCE no longer forces M/X bits set
  Now always keeping 65c816 flags in a register during opcode execution
  More source reorganization
 0.25b4 [TRAC]
  Fixed a bug in scrolling hardware ports
  Rewrote more opcodes to take better advantage of new opcode fetcher
 0.25b3 [TRAC]
  Implemented alternate 65c816 opcode fetcher (should be considerably
   faster - original one likely to be removed in the future)
  Implemented interleaved HiROM support in ROM loader
 0.25b2 [TRAC]
  Finished hooking in last of new horizontal calc code
  Slight speed-up tweak in 65c816 emulation
  Fixed a nasty bug that showed up in 0.25b1 that caused severe gfx
   corruption in at least one ROM (ActRaiser)
 0.25b1 (doh! date in readme said 4-3, should have said 23-3) [TRAC]
  Rewrote horizontal scrolling calc code to reuse data
  Reworked some of the tile cache invalidate tables/recaching code to
   store information on certain special cases of tiles to limit some
   unnecessary checks
  Implemented OAM decode caching - this increases sprite rendering speed
   noticably - more optimizations possible here...
 0.17b7 [TRAC]
  Fixed data store method of double write registers (BG scroll, M7 matrix)
  More speed up tweaks in memory mapper and screen core
  Fixed a bug in HDMA code - fixes GPFs after fights in FF5, extra
   sprites in TMNT4 and missing sprites in Killer Instinct
  Disabled DJGPP LFN support (was causing problems in the Load ROM file
   selector in the GUI, because it's still expecting 8.3 filenames)
   Thanks to Lanice for pointing out the crashing! (doh)
  Rewrote a lot more bits of the GUI to use OOP and allow more advanced
   features to be implemented in the future
  Compiling in two new fonts: the font from ZSNES (thanks zsKnight!),
   and a slightly cosmetically modified version of the same
   (only the modified font is available in the GUI at this time)
 0.17b6 [TRAC]
  More internal work on GUI
  Fixed a bug in mode 1 background priorities with BG3 high priority
   enabled - helps intro in FF5
  Oops! Removed some code that shouldn't have been in b5 - was causing
   the ROM loader to create a test file the size of the loaded ROM
  Fixed a bug that caused the color subtraction in back area hack to
   kick in sometimes when it shouldn't
 0.17b5 [TRAC]
  Fixed a bug in the ROM loader that caused it to get in an endless loop
   on ROMs that weren't an even multiple of 32k (excluding header)
  Fixed a bug that prevented HiROM SRAM from working AND allowed it to
   overwrite other memory
  Implemented SRAM mirroring in HiROM
  Reduced SRAM allocated to 64k and SRAM supported in HiROM to 64k
  Changed the value certain areas of RAM were initialized to to closer
   match SNES '9X
  Changed memory mapper so LoROM/HiROM shared more code/data
  ROM loader: replaced the fstream code with stdio code - this seems to
   fix the problem with 32k ROMs not loading correctly
 0.17b4 [TRAC]
  More work on background plotting speed, hw port emulation speed/accuracy
  SPC disabled - stopped working sometime between b2/b3 and needs to be
   rewritten anyway
  More CPU speed
  Fixed many non-emulation related bugs that caused crashes in pure DOS,
   and forced the switch to WDOSX before
  Reverted to standard DJGPP stub now that WDOSX is no longer necessary
  More fixes to LoROM SRAM mapping - SNES '9X source and one problematic
   ROM provided insight into the workings of SRAM mirroring
  Improved emulation of CGDATA/CGDATAREAD:
   now use seperate vars for odd byte reads
  DMA modes 2+ had been disabled for some strange reason (doh!) - fixed
 0.17b3 [TRAC]
  Now using WDOSX, because of having too many problems with the standard
   DJGPP stub
  Moved plenty of code around in gfx engine for potential speed increase
   and code size reduction
  Most data moved around in hw port emulation
  Most hw ports now reset in Reset (used in 'Reset Emulation' and
   'Load ROM' in GUI) - some still remain (especially in SPC)
  More work on ROM loader
  Can now go in GUI without specifying ROM on command line, however trying
   to load a ROM in GUI is still painful!
 0.17b2 [TRAC]
  Simplified hardware ports in memory mapper
  Improved emulation of OPHCT/OPVCT
  Statically allocating SRAM, WRAM, VRAM, SPC address space, OAM now
  Improved emulation of OAMDATA/OAMDATAREAD:
   now use seperate vars for odd byte reads
  Fixed a bug where 8 bytes too few were being allocated for OAM
  SRAM allocation halved (only 128k now)
  LoROM: SRAM supported reduced to 64k, writes prevented if ROM says less
   exists (allows Puzzle Bobble to start)
 0.17b1 [TRAC]
  Implemented speed throttling (thanks to Richard Mitton of the New
   Zealand Story Emulator for help with this!)
  GUI overhaul begun ('3D' appearance added, code is becoming more object
   oriented)
  CFG handler rewrite begun
  Source cleanup/overhaul begun
  LoROM 16+Mbit and HiROM memory maps fixed (thanks to zsKnight for help!)
  Implemented Hblank emulation/CPU timing (thanks to zsKnight and the '9X
   source for help and info!)
  Fixed emulation of SLHV register (thanks to '9X source for info!)
  Fixed SPC memory map to only have 32k of RAM
  Fixed a few CPU opcodes
  Fixed a memory overwrite bug in a few CPU opcodes which caused the
   emulator to deconfigure controller keymaps

 0.16f: Various bug fixes and speed-ups.

