daid / gbemulatorshootout Goto Github PK

Spotted in the build log here: https://github.com/daid/GBEmulatorShootout/actions/runs/3210664275/jobs/5248281448

Downloading https://sourceforge.net/projects/vba/files/latest/download
Downloading https://gbdev.gg8.se/files/roms/bootroms/sgb_boot.bin
Downloading https://gbdev.gg8.se/files/roms/bootroms/cgb_boot.bin
Downloading https://gbdev.gg8.se/files/roms/bootroms/dmg_boot.bin
    self._execute_child(args, executable, preexec_fn, close_fds,
  File "C:\hostedtoolcache\windows\Python\3.10.7\x64\lib\subprocess.py", line 1438, in _execute_child
    hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the file specified
Error: Process completed with exit code 1.

https://github.com/RossMeikleham/PlutoBoy

This one stands out to me as being an emulator targetting the 3DS/Vita which also supports Windows (so could be included in the Shootout).

Don't know if this is the right place/way to ask for it, feel free to shut down the request if it's not appropriate.

I'm currently developing DocBoy, a C++ GB emulator (DMG only by now) focused on accuracy.

At the moment it passes all the test roms of this repository it can support (i.e. no audio and GBC roms: I'm going to implement these soon), that is 155/263 (it passes also all the mealybug's roms and the other hard ones).

I've already forked the repo and added DocBoy (fork here) and I have GitHub's actions for automatic deployment of new releases (including Windows build, that is the one downloaded and used by my fork of the repo).

Any chance this could be included among the other emulators?

Thanks, and great job with this repo!

Requesting that you include my ROM whichboot.gb as an info ROM. It's a ROM which...

• reports the initial state of the CPU registers.
• extracts the hidden fractional bits of the DIV register.
• check the state of VRAM to see if the Nintendo logo is present.
• tries to guess which hardware or emulator it's running on based on the above.

https://github.com/nitro2k01/whichboot.gb

I went ahead and tried to implement the behavior the test was looking for into Gambatte (it was extremely easy to implement). A problem arised however, this behavior contradicts results from Gambatte's test suite.

I understand that this behavior isn't consistent between DMGs, although Gambatte doesn't seem to share the result bgb (considered "passing") has. It seems to boil down to 3 behaviors that can occur for a pixel.

1. Old BGP is read (bgb behavior)
2. Bitwise OR between old and new BGP is read (SameBoy behavior/what this test expects)
3. New BGP is read (Gambatte behavior/what Gambatte's test suite expects)

3 also appears to occur in CGB-DMG (tested on my own GB Player, and SameBoy also agrees this should be the case). However, bgb does not appear to do 3 in its CGB-DMG mode, instead still doing 1. This could maybe point to bgb instead actually having an off by one issue and actually doing 3 (but 1 is what appears to happen due to this off by one), but it could perhaps be revision specific.

Attaching one of Gambatte's testroms (there's several that fail, this is just one of them), the image that it expects, and a GBI screenshot (which matches). This testrom was also largely used for figuring out these differences. dmgpalette_during_m3.zip

It might be best overall to just mark that test as "info" due to the multiple valid behaviors.

It has a GameBoy emulator core! Technically.

I'd do it myself, but the project seems highly Windows-specific, and I don't have a Windows machine :-(

At the moment ax6 RTC rooms (ax6/rtc3test-1.gb, ax6/rtc3test-2.gb, ax6/rtc3test-3.gb) require CGB to run, but they can actually work also on DMG.

I think we can remove model=CGB for these and use DMG screenshots instead, so that they can run also on emulators not supporting GCB. What do you think?

Windesync-validate is a freshly released test ROM for a recently discovered quirk in pre-GBC hardware that can insert a glitch pixel in the pixel FIFO under some circumstances. The output from the emulator should be compared to the included windesync-reference-sgb.png. Since this is a pre-GBC quirk, the test ROM should ideally run in DMG mode.

https://github.com/nitro2k01/little-things-gb/tree/main/windesync-validate

bgb 1.5.10 was released yesterday (2022-10-18), implementing (among other things) the OAM bug

Under what license is the code in this repo provided?

As of today, I'm seeing Goomba 48/232, No$gmb 42/236, KiGB 34/231, VisualBoyAdvance 1/1. That can't be right. All tests show "No result" for VBA except daid/​stop_instr.gb (GBC). Related to #13 fix in any way?

For some reason this particular ROM is missing from the test set. Also (less importantly) blargg is misspelled with only one g.