Current version

v1.10.4 (stable)


Main page
Archived news
Plugin SDK
Knowledge base
Contact info
Other projects


Blog Archive

Altirra 1.3 released

I've released a new version of Altirra, my Atari 8-bit computer emulator. Version 1.3 contains a large number of compatibility fixes and runs a number of titles that previously blew up under 1.2. I'd like to thank Mclane and breaker for an astounding amount of compatibility testing in a comment thread of epic proportions, which tracked down many, many games and programs that did not run correctly and helped immensely in identifying and ironing out bugs.

One of the weirder aspects of doing work on this project is that it's probably the closest I've come to doing Test Driven Development. TDD is not something I'm generally sold on because I think there are lots of types of modules for which tests are not the appropriate way to ground development. In this case, though, I need to conform to a behavior that is (a) rigidly defined and testable and (b) frequently unknown. What I often end up doing is writing a unit test to determine the behavior on a real 800XL, and then tweaking the emulation code until the test passes. I have to do things this way because the emulation code is now operating at a level where it's very easy to break some programs while trying to fix others since some programs are sensitive to single-cycle timing deviations.

Another thing that's becoming abundantly clear is how difficult it would have been for Atari to create a more powerful machine with a high level of backwards compatibility. I've been finding an increasing number of programs that happen to work but have extremely tight timing margins or simply have outright bugs. I've seen programs that:

So, if you're wondering why your powerful next-gen console can't reliably run your last-gen games, now you have some idea why. When you have stable and consistent hardware, it's easier for people to unintentially write code that relies on unspecified behavior to an extreme level.

Oh, and I'll say it again: cassette tape is vile. I hated it growing up and I hate it even more now that I'm trying to emulate it. You should offer thanks every night that you no longer have to wait minutes for a program to load off of tape.

Anyway, here's the list of changes since version 1.2:

Version 1.3 [September 4, 2009]:
   [features added]
   * Debugger: Register pane now shows 65C816 state.
   * Debugger: Added context menu to disassembly pane.
   * Debugger: *.lst and *.lab symbols are automatically detected and loaded
for direct-run EXEs and cartridges. * Debugger: Improved source-level debugging capability. * Debugger: Opcodes are now cached so that they show correctly in the history
log even with bank switching. * UI: Filter mode is now saved on exit. * UI: Added command to copy current frame to the clipboard. * UI: Added paddle support. * GTIA: Optimized artifacting code. * GTIA: Added the ability to disable collisions for cheating. * Cartridge: Added support for 64K XEGS and 512K MegaCart cartridge types. * Cartridge: Added support for SuperCharger 3D cartridge. * Cassette: Added support for non-standard baud rates. * Cassette: Added auto-boot option. * Cassette: Device open commands are now accelerated. [bugs fixed] * Simulator: More fixes to executable loader. * Simulator: 320K memory mode now banks ANTIC along with the CPU. * Simulator: Joystick and bank switch masking via the PIA data direction
register is now supported. * Simulator: Rewrote BASIC handling to use the OPTION switch rather than
masking PORTB bit 1. * Simulator: Reset PIA, ANTIC, and CPU on warm reset for XL hardware. * Simulator: Self-test ROM no longer appears if kernel ROM is disabled. * Simulator: Adjusted PAL frame rate. * Simulator: Disk boot flag is now set when loading EXEs to accommodate
programs that warm start through DOSINI vector. * Simulator: EXE loader is more tolerant of broken files. * CPU: Fixed many 65C02/65C816 bugs. * CPU: Fixed emulation of NMOS 6502 decimal mode. * CPU: Fixed bugs in undocumented instructions. * CPU: Fixed emulation crash when switching to 65C816 on the fly. * CPU: Fixed handling of break flag. * CPU: Now allow one instruction to pass before taking IRQ interrupt after
clearing I flag. * Debugger: Improved handling of 65C816 modes in disassembler. * Debugger: Debug memory reads no longer see $FF in the page where a read
memory breakpoint is set. * Debugger: Fixed bugs in symbol parser when loading MADS listings. * Debugger: Fixed line highlighting in source windows. * Debugger: Disassembly recomputes on 65C816 mode changes. * Disk: Fixed bug in loading of ATR files that have non-standard boot sector
counts. * Disk: Additional compatibility fixes to DSKINV acceleration. * Disk: Status commands are no longer delayed by the accurate disk timing
option. * Disk: Applied slight adjustment to SIO transfer rate. * Disk: POKEY timers are now shut off after SIO/DSKINV acceleration. * Disk: Strengthened disk image format validation and made XFD loading more
tolerant. * UI: Fixed bug in CPU mode dialog that prevented selection of 65C02 mode. * UI: Fixed crash when accessing menu with display pane closed. * UI: Keyboard-as-joystick logic no longer allows simultaneous up+down and
left+right. * UI: Added support for controlling joystick port 2. * UI: Blocked invalid machine configurations with XL kernel and 800 hardware or
less than 64K of memory. * ANTIC: Fixed broken mode 9. * ANTIC: Vertical reflect now works on modes 4-7. * ANTIC: Player/missile DMA no longer occurs during vertical blank. * ANTIC: Fixed behavior of vertical scrolling on blank lines. * ANTIC: Changes in playfield width are now supported later in a scan line. * ANTIC: Moved VCOUNT advance forward by one cycle. * ANTIC: NMIs are now suppressed if enabled too close to trigger point. * ANTIC: NMIs are now delayed by one cycle if enabled exactly on cycle 8. * GTIA: Bogus missile-player collisions were reported if a register change
occurred in the middle of a missile. * GTIA: Fixed CONSOL I/O interactions on switch lines. * GTIA: Save states now restore player/missile widths. * HLE: DSKINV now supports double-density disks. * HLE: Improved compatibility of post-init PORTB state. * POKEY: Rewrote resampling filter to better block high frequencies. * POKEY: Fixed 17-bit noise register polynomial and shift direction. * POKEY: Initialization mode (SKCTL[1:0] == 0) now resets the low-speed clocks
and polynomial shift registers. * POKEY: Asynchronous serial receive mode no longer suspends channels 3 and 4.


This blog was originally open for comments when this entry was first posted, but was later closed and then removed due to spam and after a migration away from the original blog software. Unfortunately, it would have been a lot of work to reformat the comments to republish them. The author thanks everyone who posted comments and added to the discussion.