Emu#dreams Forum | Download | RSS | IRC | ADD NEWS | last 7 days | last 31 days | MUSEUM

Altirra 3.90 Test5 [141][witek], 2019-07-26 14:19:12

[Source: https://atariage.com/forums/topic/292556-altirra-320-released/page/5/#comments]

test 4

- Some latency reduction work. The split for frames has been moved from scanline 0 to scanline 248, since anything that depends upon the OS for input will sample controller input during VBI processing. Per-frame controller processing has also been reordered to reduce latency for certain types of emulated controllers, particularly light pens. Should knock off a frame of delay for some cases.
- Fixed excessive polling of host gamepads in vertical blank.
- PIA port A (controllers 1+2) is now included in performance traces.
- Added information to the help file about bank-sensitive cartridge debugging, and fixed a bug with line number information not being fixed up in this mode.
- Replaced /dark command-line switch with an official option in the UI, replaced some standard message boxes with the newer style, and fixed up the memory and source windows to be dark-aware.
- History window go-to-source and go-to-disassembly are a bit more aware of global addresses now, go-to-source now reports when the source lookup fails, and go-to-disasm can be done via double-click.
- AltirraOS 3.21: KRPDEL implemented and keyboard repeat rate is properly adjusted for PAL.
- Fixed crash when pasting into enhanced video display in CIO intercept mode, and clear attract mode counter when accepting input.
- Initial version of trace cache for 6502 coprocessors used in drive emulation. This provides a minor speed boost and is a prerequisite for JIT, but should not have visible effects otherwise (knock on wood). This is only enabled for plain 810 and 1050 drives as it does not support ROM banking.
- Fixed cycle count bugs in the several illegal 6502 instructions, ARR #imm and read-modify-write instructions using indexed modes. The RMW instructions were not forcing the extra indexing cycle as is usual for store instructions. Normal instructions werent affected and I im not aware of any software that broke due to this -- was only found in a test.
- The crash handler now reports if it couldnt write the dump file.
- The emulator now falls back to the "you are missing something" ROM if it couldnt load the OS ROM file.
- Fixed broken option to auto-load kernel ROM symbols.
- Added option to suppress automatic loading of system symbols (OS database, entry points, and hardware). They can still be loaded manually with .loadsym if desired.
- Disassembly window does some checks to see if memory has changed, to reduce showing stale code when doing a fast update.

test 5

- Fixes crash in Performance Analyzer with system display scale >100%.