Nowy test emulatora AMIGI - WinUAE, autorstwa Toni Wilena.
Nowy test emulatora AMIGI - WinUAE, autorstwa Toni Wilena.
Winuae 3.3.0 Beta 17 30/05/16
Beta 17: (Perhaps last or second to last beta)
- Pre-KS1.3 boot hack code and data was overwritten by new resident structure update in b4.
- PlanarToDirect RTG optimization fixed. (b16 included forgotten borken version).
Winuae 3.3.0 Beta 16 28/05/16
- Replaced obsolete Logitech LCD API usage with current version. No other changes, still plain black and white only.
- Fixed IDE emulation corner case when polling (not using interrupt) BSY bit.
- Changing GUI relative path mode immediately updates paths stored in registry. (Older WinUAE versions don't understand relative paths in registry, switch off relative mode if you use old versions)
- Fixed directory filesystem slowdown introduced when it was converted to OS4 compatible trap system (b3).
- Replaced obsoleted _wpgmptr with GetModuleFileName()
→ NOWSZY [AMIGA] Winuae 3.3.0 beta 18
Winuae 3.3.0 Beta 18 03/06/16
Beta 18: (another possible last beta)
- "ROM needs 68020+" dialog repeated forever.
- Fall back to old GDI screenshot method if GetRenderTarget() fails.
- Logitech color LCDs work again, old API did the conversion automatically.
- Fixed UAE boot ROM trap code corruption if "too many" UAE expansions were enabled. (b3)
→ [AMIGA] Winuae 3.3.0 beta 15
Winuae 3.3.0 Beta 15
- Remove unmatched characters from keyboard injected paste event. (Previously was replaced with spaces).
- Delay each paste injected key press and release a bit. Some programs don't like too fast key events.
- Restoring statefile with blitter active and it was force-finished (=not cycle-exact mode): blitter interrupt was not correctly generated when restoring state file.
- Fixed Direct3D mode DirectX9 install warning buffer overflow.
- Added "ROM disabled" to ROM panel Boot ROM selection. "ROM disabled" = force disable UAE boot ROM even if one or more UAE expansions are configured.
→ [AMIGA] Winuae 3.3.0 beta 14
Winuae 3.3.0 Beta 14
- AGA subpixel scrolling: if line's parameters didn't change compared to previous field's line, line was not marked as "changed" and was not always redrawn correctly. (Roketz scoreboard corruption)
- CD32: always generate subchannel interrupts and transfer subchannel frames when reading CD sectors if subchannel DMA channel is enabled. (Previously was only done when playing CD audio) Real hardware also transfer subchannel frames when CD is idle. This is not yet emulated.
- Added "Paste" input event (END+INSERT default mapping), injects host clipboard text content using emulated keyboard key codes. Only available when clipboard sharing is not active and only supports ROM default US keyboard layout. For example can be used to paste strings directly to boot CLI, hardware debuggers or non-AmigaOS operating systems.
- Mouse/joystick autoswitching broke some betas ago.
- Pause/GUI + black frame insertion: don't show black frame.
- IVS Trumpcard Pro/GrandSlam SCSI emulation implemented.
- AROS ROM updated.
- Few weeks ago released new MSVC optimizer enabled (May not have it enabled in final version because it is not yet fully production ready). Report if you notice any performance improvements.
IVS Trumpcard Pro/GrandSlam SCSI:
- Autoboot, seems to support custom filesystems.
- 5380 based.
- Fake DMA. Long reads, word wide writes.
- "Trumpcard Professional Driver 4.9 - devs/IVS_SCSIpro.driver IVSSCSI 4.9 (20 Sep 1991)"
→ [AMIGA] Winuae 3.3.0 beta 13
Winuae 3.3.0 Beta 13
- HALT10 was incorrectly triggered in some configurations (b12)
- Update ExecBase.MaxLocMem (and ChkSum) if more than 2M chip RAM is configured.
- It was not possible to fully disable enabled expansion device without restart. Checkbox was unticked but device was not fully disabled.
- Copper COPJMP blitter conflict bug emulation is again disabled (Breaks Batman Vuelve / Batman Group). There is still something unknown. I did some more tests and some blitter cycle sequences do not conflict with buggy copper cycle and correctly wait one extra cycle.
- Added IVS Trumpcard Pro/GrandSlam SCSI v4.9 ROM image. Emulation not yet implemented. "Trumpcard Professional Driver 4.9 - devs/IVS_SCSIpro.driver IVSSCSI 4.9 (20 Sep 1991)"'
Variable Refresh rate updates. I finally bought G-Sync monitor, Acer Predator X34. Every emulator (any emulator!) user should get G-Sync or FreeSync hardware because variable sync has all the advantages of vsync without any vsync disadvantages!)
- Adaptive Sync renamed to Variable Sync. Adaptive Sync is name of DP variable refresh rate standard (which AMD's FreeSync is based on). "Official" generic name is Variable Refresh Rates but Variable Sync fits better with other VSync modes in GUI.
- Variable sync + black frame insertion is now supported in windowed and full-window mode. (I couldn't get G-Sync to work in normal windowed mode but it did work in full-window)
- Variable sync + black frame insertion is now supported in RTG modes.
- Black frame insertion is not enabled in variable sync mode if refresh rate would become larger than selected display refresh rate (fullscreen) or desktop refresh rate (windowed/full-window).
- Variable sync + black frame insertion and if doubled refresh rate is equal to selected/max hardware supported refresh rate: automatically reduce internal emulated refresh rate by 0.5Hz to guarantee glitch free display. (Workaround for monitors that have max refresh rate of 100Hz=PAL or 120Hz=NTSC)
→ [AMIGA] Winuae 3.3.0 beta 12
Winuae 3.3.0 Beta 12.x
- winuae.7z updated, halt10 in some configurations fixed. Fixed. (Weird boot ROM code points stack at ROM when it does CPU check, I guess it is acceptable if you don't need to return from exception.)
Winuae 3.3.0 Beta 12 / 05/05/2016
- Dummy 68030 MMU instructions were also accidentally disabled in "generic" 680x0 core. (b11)
- AGA mode HAM6 colors were not 100% accurate. (Was fixed in b11)
- Stop with HALT10 if compatible CPU is not ticked and exception begins with SSP pointing to non-existing address space.
- Prevent programmed mode display width and height to become too small. Program doing something very stupid or crashing program can write very random values to programmable refresh rate custom registers.
- Switching Advanced chipset compatible option off and then back on does not anymore reset chipset extra back to Generic in non-quickstart mode.
- 1M/512k+512k Advanced chipset option was not reset when Advanced chipset was in compatibility mode.
- A3000 SCSI + any WD based Z2 SCSI controller: autoconfig didn't work.
→ [AMIGA] Winuae 3.3.0 beta 11
Winuae 3.3.0 Beta 11 / 24/04/2016
- Screenshot/capture without "before filtering" ticked now uses much faster GetRenderTargetData() function in D3D mode to capture the output.
- Ticked 68060 unimplemented CPU emu checkbox also enabled 68030 MMU instructions.
- Terminate indirect UAE Boot ROM trap threads before unmapping Amiga side memory to prevent random crashes when exiting emulator with traps active.
- Debugger il mask is now accepted in two 32-bit parts, 64-bit values are not supported by debugger input.
- Added gfx_black_frame_insertion_ratio config entry, sets timing between normal and black frames in adaptive sync mode. Default is 50 (50%, normal and black frames have same duration).
- CD32 emulation updates, mostly previously unknown undocumented features:
- It seems unmapped addresses don't "float", all invalid reads seem to always return zeros.
- CIA chip select, address bits 12 and 13 are CIA selects. Gary: 0=none,1=A,2=B,3=A+B, Gayle: 0=none,1=A,2=B,3=none, Akiko: 0=A,1=A,2=B,3=B (Very unexpected)
- CIA address space: Gary: $A00000-$BFFFFF, Gayle: $BFDxxx and $BFExxx, Akiko: $BFE000-$BFFFFF.
- Custom register mirror at $B90000-$B9FFFF! (This is really weird..)
- Usual custom register mirror at $C00000-$CFFFFF.
- Akiko addresses are mapped from $B80000 to $B87FFF (was $B8FFFF) and has 64 byte mirroring (was no mirroring)
- Akiko ID at $B80000.l is $C0CACAFE (was only $CAFE at $B80002.w, KS checks this address for $CAFE)
- All write-only registers seem to read same data as nearby read-only register.
- Interrupt registers only have bits 24 to 31 writable, other bits always read as zeros.
- Config ($B80024) register has bits 23 to 31 writable, other bits always read as zeros.
- Subchannel arrived interrupt bit is set at boot for some unknown reason, subchannel index register ($B80018) works strangely and has unexpected value at boot ($C2).
- Match Akiko C2P behavior with real hardware when reading bytes/words or when doing multiple read passes.
- When booted with CD inserted: first packet from CD MCU is always "media inserted" packet.
- $B80028.b reads last received command packet byte.
- Writing to $B80028.b sends command bytes to CD MCU. (This is not emulated yet and no program uses it)
- Akiko internal CIAs don't have external TOD input pins. CIA-A TOD which normally counts vsyncs or power supply ticks count rate is selected with $B80020 bit 23 (0=50Hz, 1=60Hz). CIA-B (hsync) timing logic is not known yet but it is also internally generated.
- Old interrupt request set/clear hacks removed.
- All CD status return codes (including errors) include door status (open/closed) in bit 0.
- Contents of unused TOC entry bytes now contain real hardware matching values.
→ [AMIGA] Winuae 3.3.0 beta 10
Winuae 3.3.0 Beta 10 / 10/04/2016
- if uaenet.device is online and packet(s) have arrived but there is no pending CMD_READ/CMD_READORPHANs: wait about 2s before dropping it, it is possible that multiple packets will arrive so quickly that Amiga-side TCP/IP stack don't have time to queue more read commands before packets get dropped.
- uaenet.device stability fix.
- Loading config with unplugged joystick or mouse inserted in either port: previous device was not autoselected.
- Filter panel scaling mode active and mouse driver enabled: mouse got stuck in a corner in some situations.
- Screenshot input event set to toggle-mode in Input panel: key/button press = create screenshots continuously every frame, following key/button press = stop automatic screenshots.
- one genlinetoscr function (16-bit 4x horizontal stretch) used genlock function in non-genlock mode.
- Adaptive Sync "vsync" mode added. Optimal setting designed for G-Sync/FreeSync monitors. Also allows 100/120Hz black frame insertion mode. Not very stable yet. (Don't try to use without adaptive sync hardware..)
- 68040 MMU mode SR M-bit handling was missing. 68060 SR M-bit handling was not correct.
- Disable bsdsocket.library if indirect UAE Boot ROM is enabled, it is not (yet?) compatible.
WinUAE Beta 9 19 March 2016
- Fixed swapped audio channels if stereo separation was set (b1 cleanup missed code path)
- Fixed uaehf.device SCSI emulation sense data possible crash/wrong status. (trap update related)
- 32M UAE RTG board was always address space aligned. It is wrong in UAE mapping mode.
- Only swallow hardware horizontal and vertical mouse movement when in mouse driver mode, not mouse wheel.
- Improved programmed refresh rate hsync start and end validation to prevent crashes if weird mode is set. (BoggleDop / Sanity)
- Added hardfile virtual bad block support. uaehf.device and hardware SCSI emulation only. Create geometry text file (<name of hardfile.hdf>.geo) in same directory as hardfile, add badblocks=<blocks> line. (<blocks> can be range, 10-100 or single value or multiple ranges/values, for example 10-100,200,1000-20000) Max 8 bad block values/ranges supported. Error is returned if block in badblock range is attempted to be read or written.
- OS4.x UAE trap mode now uses hybrid busy wait mode. Performance has improved again, directory filesystem speed is now fully usable and uaegfx is also noticeably faster.
- UAE indirect trap stability updates.
WinUAE Beta 8 05 March 2016
- Game Ports joystick type was always empty (undefined).
- Quickstart panel CDTV/CD32 CD path name label was "DF1:", not "CD". (Broke when bridgeboard drive support was added)
- Dragndrop to emulation window didn't work correctly until GUI was opened at least once.
- Added screenshot autoclip mode, uses autoscale display size/position detection. Also added following config file only entries: (all config entries need win32. prefix)
- screenshot_min_width and screenshot_min_height: minimum autoclip size in lores pixels.
- screenshot_max_width and screenshot_max_height: maximum autoclip size in lores pixels.
- screenshot_width and screenshot_height: forced screenshot size (non-autoclip mode only)
- screenshot_x and screenshot_y: position/offset of screenshot (non-autoclip mode only)
- screenshot_mult_width and screenshot_mult_height: 1x, 2x, 3x, .. 8x screenshot size multiplier.
- screenshot_output_width_limit and screenshot_output_height_limit: minimum screenshot size, if original size is smaller, multiplier is increased until width/height is same or larger.
- More internal changes to support more than one enabled RTG board at the same time. (Still lots more to do..)
- 68020/030 RTE didn't generate exception when it encountered 68040+ or 68010 only stack frame types.
- Return error code instead of crashing if bsdsocket emulation sendto() tries to use unsupported raw protocol.
- Switching from relative path mode back to normal mode could have caused repeated crashing.
- Do not write reserved block at the end of FileInfoBlock back to Amiga side, Pre-2.x dos commands don't allocate complete FIB structure. (Was side-effect of OS4.x compatible trap system)
- Basic NERO nrg CD image support added. Data and audio tracks seems to work.
- CDFS mounting selects Joliet volume label if found. (Joliet volume label is not restricted to capital ASCII letters)
- uaenet.device S2_READORPHAN should not filter unknown multicast addresses.
WinUAE Beta 7 21 February 2016
- Fullscreen + paused + enter GUI: GUI was invisible.
- OSD FPS counter shows 'P' in pause mode.
- Directory filesystem drive total/used values are not anymore converted to internal block count before again converting to AmigaOS block count. Reported Amiga-side size is now accurate.
- Create more sane looking fake geometry for directory harddrives. (Not that anyone really cares)
- Directory filesystem harddrive block size is dynamically adjusted to enable correct drive size information even if drive is large (>2T), as long as AOS also supports it (OS3.9+ probably required).
- If m68k FPU is in float (32-bit) precision mode, force output register value to 32-bit precision after each host FPU operation to guarantee correct result. (Kilofix / Iris missing geometry)
- Clear UAE boot ROM temporary variables at reset. (OS4.x reset loop after warm reset)
- Added 20 bytes of empty space after directory filesystem volume node. If OS4 decides to use it for something. (OS4 volume node is 20 bytes larger. UAE puts it in hardware board space, it can't use DOS functions to allocate it).
- Implemented OS4 ACTION_FILESYSTEM_ATTR directory filesystem dos packet.
- Selecting none in Game Ports reverted back to previously selected device/layout.
WinUAE Beta 6 14 February 2016
- Fixed uaehf.device and input on the fly device change invalid free() calls.
- Game Ports panel autofire/joystick type setting was not remembered.
- Do not queue new uaenet.device uae_Signal() call if previous is still in queue.
- Do not call any host OS functions from uaegfx trap functions, trap function can run in different thread context which can cause strange side-effects. OS4.x uaegfx resolution changes should be safe again.
- Added shellexec custom input event/uae-configuration command that can execute Amiga-side commands or scripts. Uses SystemTagList() with NIL: Input and Output handles if KS2+, Execute() with NIL: Output if pre-KS2. Only works when UAE Boot ROM is active, dos.library has been initialized and AmigaOS is running normally.
- Display panel color component gamma value calculation fixed, any brightness or contrast modification was too dark.
WinUAE Beta 5:10 February 2016
- Game Ports mouse and joystick was always reset to defaults (b4)
- Dark palette fix adjustment is now calculated before other adjustments (brightness, contrast, gamma)
- Dark palette fix third option added, pre-converted Atari ST palette increased to full range, 12-bit EEE becomes nearly full white, FCFCFC in 24-bit RGB space. (Not FFFFFF because palette would not be linear anymore)
- OS4 UAE expansion updates: (Discussion thread here: http://eab.abime.net/showthread.php?t=81146)
- Directory filesystem random errors fixed.
- Directory filesystem/CDFS CD on the fly insertion/removal supported! (It wasn't that difficult after all)
- uaescsi.device supported.
NOTE: Try not to change uaegfx resolution in OS4, it can cause mysterious crashes or side-effects. Fix coming later.
TODO: bsdsocket.library and fixing random hangs.
WinUAE Beta 4 07 February 2016
- Default game ports config without default.uae (Mouse + keyboard layout A) was not set. (b3)
- On the fly input device change crash fix. (b3)
- Monochrome mode config file entry added.
- Atari ST dark palette fix moved to Brightness/Contrast/etc select menu and added alternate mode.
- UAE expansion resident structures are now injected to execbase ResList. Now less important expansions are initialized later. Previously all were initialized during diag init time. Now also uses normal RTF_AFTERDOS resident to start clipboard sharing instead of ugly hack.
- Simplified rawinput support, also removed rawinput checks, added when long time ago WINE didn't fully support rawinput.
- Added harddrive master write protection checkbox to Misc panel. WARNING: don't use it with hardfiles unless you know exactly what you are doing.
- UAE Boot ROM with PPC native OS message now asks to enable correct boot ROM mode.
OS4 UAE expansion updates: (Use this thread: http://eab.abime.net/showthread.php?t=81146 for this topic!)
- Clipboard sharing is now supported.
- Magic mouse is now supported.
- Virtual mouse driver is now supported.
- uaenative.library converted to new trap system but not tested.
- uaescsi.device initialization converted to new trap system, device will open now without crashing but any CD access commands will not work (and most likely will also crash strangely).
Still not supported:
- Directory filesystem on the fly insertion/removal, including CDFS CD swaps. Getting this to work will be extremely complex task. Not sure if it is worth the trouble. Unless I get lots of donations
- uaeahi (Probably not needed, PCI sound card emulation is better choice)
WinUAE Beta 3 31 January 2016, 17:46
- It says b2 but just ignore it for now...
- When matching config file and on the fly connected game controllers, first check if both id and friendly name match and only if no match, check id only and finally name only. Some multi slot controller adapters have unique name for each port but all ports have same ID. (Was id and then name matching previously)
- Remembered input device selections are now forgotten when new config is loaded.
- On the fly unplugged devices are now properly remembered and re-inserted if on the fly reconnected.
- PC Bridge disk drives raw image support (extended adf, ipf and so on.) Standard DOS disks only, no copy protections or other tricks supported. Read-only, writing is not supported.
- JIT slider now only shows power of 2 cache size values and max is 16384. (Max was already changed in previous betas)
- Linked config files loaded unreliably (or never) due to uninitialized variable.
- Real storage device (harddisk, memory card etc..) failed to open the device if it didn't have device path, only drive letter.
- CDFS automount didn't mount CDs with empty label, dos does not like empty volume names, replace empty name with "NO_LABEL"
- BPLCON0 BYPASS bit emulated. AGA-only, bypasses palette selection, creates gray scale output. (Microcosm cheat mode effect)
- Added A1000 composite out monochrome output mode emulation when BPLCON0 COLOR bit is zeroed. Option in advanced chipset panel. Only emulated if no video port display devices enabled, monochrome emulation shares same code (and in real world it is also not possible, RGB video out is not affected by COLOR bit). Also added always-on monochrome
- mode option to Display panel. (Uses HRM documented gray scale brightness = 30% red, 60% green and 10% blue)
- "Atari ST palette fix" added to Display panel, called "Dark palette fix". It is wrong to mention Atari ST in GUI There are few stupid Atari ST ports that use original palette (3 bits/component) = halved brightness in Amiga.
- UAE autoconfig board/boot ROM selection added to ROM panel. Do not touch unless you wan to run OS4 + UAE expansions. OS4 + UAE expansions: must be set to last option, "New UAE (128k, ROM, Indirect)". (Quickly added to GUI, probably will change in future versions)
OS4 compatible UAE autoconfig board, boot ROM and communication interface.
OS4 UAE expansion thread is here, all OS4 related goes there:
- Directory harddrives.
- uaehf.device (UAE HD controller hardfiles)
- uaegfx RTG. (Copy OS3.x uaegfx and uaegfx.info to OS4.x Devs:Monitors)
- uaelib ("uaelib_demux" trap)
- bsdsocket.library (This is the most difficult)
- clipboard sharing
- virtual mouse driver
- on the fly directory harddrive changes.
- Slower than normal host<>Amiga communication because every read or write that accesses Amiga memory needs to be done in Amiga side (due to non-1:1 logical/physical mapping). It will never be as fast as direct communication.
- Remember to set ROM panel UAE expansion type to "New UAE (128k, ROM, Indirect)".
- Directory harddrive (copied from HDF) OS4 installation will boot but it is not recommended, it is slower than HDF and there is mysterious issue that causes random read errors. (Missing font requester when booting is the most common side-effect)
- Most UAE expansions have preliminary updates for indirect trap support and support required multiple major changes = many things can be strangely broken..
Development speed of OS4 compatible UAE expansion and priority of features mainly depends on donations.
→ [AMIGA] Winuae 3.3.0 beta 1
Winuae 3.3.0 Beta 1 / 09/01/2016
- Added extra validation checks for STORAGE_DEVICE_DESCRIPTOR, buggy drivers may return invalid (negative) offset values.
- 64-bit JIT harmless "Warning! 12 is locked" log spamming disabled.
- Updated copper/blitter cycle conflict emulation. It only triggers when CPU writes to COPJMP, Copper is waiting and Copper DMA is enabled (and blitter uses same cycle). If Copper DMA was off, bug won't trigger, even if DMA gets
- enabled few cycles later. Copper bug emulation is again always enabled if 68000 and cycle-exact config.
- Cancel main thread sleep state immediately (Sleep due to CPU idle or similar) if PPC thread caused mainboard interrupt. This should reduce PPC to M68K interrupt emulation latency.
- Hardware RTG emulation rendered same frame twice in some situations. Caused very slow performance in triple buffered mode.
- Debugger keyboard presses were buffered and output to Amiga-side when debugger was exited.
- Starting disk DMA does not flush Paula internal 16 bit buffer. Poll DSKBYTR for DSKSYNC, immediately start disk DMA with WORDSYNC enabled: following sync word must not be missed. (Original Virus really slow loading in cycle-exact mode)
- Chip ram was marked as 16-bit Fast RAM for the CPU (no DMA contention emulation) if any memwatch point was active.
- Amithlon partition type (0x78/0x30) support was broken, partitions were detected but it was not possible to mount them.
- Added SX32 Pro board and ROM image. (Not all RAM sizes are correctly mapped)
- Only unmap Z3 Fast RAM at reset if Z3 mapping mode is "Real" for better compatibility with old configs that assume Z3 RAM does not temporarily disappear at boot.
- JIT on/off on the fly change without other CPU settings changed at the same time with uae-configuration didn't work.
- Hardware emulation autoconfig board UAE interface first working feature: uae-configuration now works under OS4 Requires also updated uae-configuration, interface can also work under other non-m68k AmigaOS operating systems, only requirement is simple native uae-configuration port. Interface design is not final yet.
- If CSPPC or BPPC is booted with empty or zero size flash rom image, fake resident idtag is automatically added that fools official flash rom updater to detect working board, "updating" the empty flash with full image (updater contains full flash rom image in encrypted format)
- Debugger memwatch points can now match only CPU instruction (CPUI) or only CPU data accesses (CPUD, CPUDR, CPUDW)
- "Heat map" debugger added, shows optional visual heatmap of CPU and DMA accesses. Can list highest addresses used by CPU instruction fetches or all addresses that single device accessed (Some DMA channel, CPU)
- Some AGA modes were incorrectly detected as bitplane DMA overrunning causing screen corruption.
- AROS ROM updated.
Input system and mapping updates:
- Added SPC_SWAPJOYPORTS input event, swaps gameports joystick ports.
- Added END+J default mapping = SPC_SWAPJOYPORTS.
- Device autoswitch didn't ignore button release without matching press.
- Device autoswitch switches to new Game Ports custom event if it has fire button mapped to mouse or gamepad.
- "WinUAE null keyboard" renamed to "WinUAE keyboard", this is now the default global virtual keyboard. GamePorts panel only uses this keyboard, multiple keyboards are only supported in Input panel mode and only if manually enabled. Reduces complexity and confusion when moving configurations between systems with different number of keyboards.
- Game Ports custom mapping rewritten. Now Game Ports list has 6 custom slots which can me inserted to any joystick port. New custom mapping data is also saved to separate config entries, one per slot.
- Old Game Ports custom config is automatically converted to new custom mapping system.
- Removed X-Arcade built-in layouts. Re-create them with new custom mapping system if needed.
- Cleared custom input event string equals setting event to "<none>".
- Fixed since the beginning bug that caused incorrectly matched input device(s) and input config when config was loaded with different connected input devices and order of devices was also different.
- Rewritten game ports validation (Detect and fix if same device in multiple ports etc..).
- Old config files should still work like previously except: game Ports custom mapping and first keyboard input config goes to WinUAE keyboard slot.
- Last few Game Ports panel inserted devices (layout or real input device) are remembered. For example if config is loaded with joypad A inserted in port 2 but pad is not plugged in, then user manually changes it to joypad B. Now if pad A is later reinserted and pad B is removed, A is automatically put in port 2. It also works with keyboard layouts or new custom layouts, if keyboard layout was previously selected, then it was later replaced with gamepad, if gamepad is unplugged on the fly, previous keyboard layout is automatically selected.