emu-russia / breaknes Goto Github PK

Сейчас:

• Каждый чип содержит enum Revision
• Revision - это просто перечисление всех вариантов исследованных чипов
• Внутри кода проверяется ревизия и симулируется или не симулируется схема, специфичная для данной ревизии

Надо:

• Оставить enum Revision, но добавить туда ещё Custom
• Добавить структуру Features для каждого чипа с boolean полями для включения отдельных "фич" (особенностей) aka симуляцию специфичных схем
• Код теперь будет проверять не Revision, а флажок в структуре Features для решения - симулировать фичу или нет
• В конструкторе чипа заполнять структуру Features для стандартных значений Revision дефолтными значениями
• Для кастомных ревизий (Revision = Custom) структура Features задаётся с помощью вызова API SetChipFeatures

Таким образом останется возможность использования как "стоковых" чипов, которые мы изучили, так и создание кастомной "химеры" со своими уникальными особенностями (схемами).

Зачем: @eugene-s-nesdev просил слепить химеру из чипов фамиклона, для создания "идеального" фамиклон-чипа (без багов).

Some small sawtooth is coming from the DPCM channel:

We need to add a tab and support for simulation breakpoints.
This is not hard to do. Just call GetDebugInfo from BreaksCore and check breakpoint condition. If it matches, display MessageBox and set Paused=true.
This way you can e.g. stop PPU on required scanline (H-Counter) or e.g. when some signal takes some value.

• Develop a metainformation schema for NES/Famicom ROM files (JSONES aka JSON NES)
• Make a Runtime converter of .nes format into internal representation of metainformation
• Popularly explain the idea in the form of a presentation

We need to add a VRAM display (Name Table + Attr Table), familiar to humans.

it's all playing at the same frequency.

By @quietust

copyright menu only

Make the sprites appear on the Debug tab.

Something strange is going on in APUSim. We need to figure out what exactly is wrong there and fix all the little bugs.
Presumably it has to do with the sequence of APU blocks simulation (propagation delay).
But we have had cases when the errors were found even in the chip topology (PPUSim). As for the APU, it must not be that bad - after all we have a working model for Logisim and an APU implementation in HDL which is covered by a large number of test benches.
The best way to catch errors would be to add signal tracing, like in PPUPlayer.

Something strange is going on with Color RAM.

It either doesn't display correctly, or it doesn't fill up correctly with values.

PCLK keeps moving away (video visualizer)

In this screenshot, Color Burst has gone to the visible part of the signal.

Once you have FamicomBoard.cpp + NESBoard.cpp you need to remove all of the copied pieces of code:

• BreakasmInterop -> BreaksCore
• M6502CoreInterop + BogusSystem.cs -> BreaksCore BogusBoard.cpp
• DataPathControl.cs -> SharpTools
• PPUPlayerInterop -> BreaksCore PPUPlayerBoard.cpp
• PPUPlayerInterop.cs -> BreaksCoreInterop.cs -> SharpTools
• LogisimHEXConv.cs -> SharpTools
• SignalPlotControl.cs -> SharpTools
• CustomClass.cs -> SharpTools
• DSound.cs -> SharpTools
• FurryPlotControl.cs -> SharpTools
• NSFPlayerInterop -> BreaksCore NSFPlayerBoard.cpp + APUPlayerBoard.cpp
• NSFPlayerInterop.cs -> BreaksCoreInterop.cs -> SharpTools

Using the reference circuit: https://github.com/ogamespec/breaks/blob/master/Docs/Famicom/Famicom_Scheme_IO_Annotated.png

• Define ports
• CPU(APU) instance
• PPU instance
• Sim SRAM chips
• Sim aux. logic chips

With V1.8, I'm seeing corrupt VRAM on the following games:
10-Yard Fight
Mach Rider (PRG0)
Soccer
Dragon Quest II
Magmax
Wild Gunman (PRG1)
Urban Champion
Zippy Race

And separately:
Pac-Land (scrolling problem)

I don't know if that helps, but I can look for more. Fascinating project.

When there are too many samples, something overflows somewhere:

But it continues to work by a miracle.

By @andkorzh

It's about time to develop a subsystem for interacting with controllers and Expansion Ports.

• BreaksCore API for IO
• Board bindings
• Famicom controllers
• NES controllers
• Update BoardDescription.json for IO
• Virtual controllers (DEBUG)
• Managed input settings dialog + ioconfig.json
• Managed InputProcessing.cs
• Check everything works

Add a basic framework for implementing APUSim. The application must be able to emulate 6502 (NSF format requirements) and play the audio buffer under .NET.

Develop an engine to simulate NES/Famicom mappers, using the 6502 microcode.

• apu.cpp
• clkgen.cpp
• core.cpp
• dma.cpp
• dpcm.cpp
• length.cpp
• noise.cpp
• square.cpp
• triangle.cpp
• debug.cpp

Soon Famicom will come to disassemble, we will see what is in the chip and make LS139(DMX) properly

Add possible infinite rendering of what is in VRAM/OAM/CRAM/CHR, without using the CPU I/F registers dump.

• Scrolling registers change independently and the picture "swims"
• The image does not match the patterns
• You can barely see the logo

Scans:

Also not very healthy, because the visible part of the signal is very sloping and rises just above the black level. Although the phase component slips here and there. Well, in fact, it can be seen on the screen - blackness + sometimes dark spots.

Apparently something is going on with the TV scroll (TV PAR Counter / SCCX TV register / something else)

• EvenOddOut EVAL
• V0 Hold
• Shifter
• Chroma Decoder
• VBL Regs
• OAM0-7
• BLNK Eval
• W3 Enabler
• VideoProcessing PAL PLL

https://www.nesdev.org/wiki/VTxx

Systems include VT02 (no extra RAM settings), VT03 (with cart SRAM enabled), VT09 (with internal RAM at $0000-$0FFF, and with cart SRAM enabled), and VT32 (Same RAM settings as VT09). maybe later for VT369 if figured out.

These settings are based on Nintendulator-NRS, so take it with a grain of salt.

We need to figure out why the wrong colors are being displayed. Check:

• Is it CRAM?
• Is this VideoOut.cpp?
• Is it an incorrect Goertzel filter?
• Did I forget to add Color Burst for Goertzel?
• Is it BGCOL not outputting correctly?
• Is it something else?

There's something strange going on at NSFPlayerBoard right now. The last time there were problems with setting the PC for the 6502 and "calling" the INIT/PLAY procedures. When all small bugs in APUSim are fixed, go back to NSF support and get this thing working properly.

https://forums.nesdev.org/viewtopic.php?p=274887#p274887
https://forums.nesdev.org/viewtopic.php?p=286245#p286245

Выяснено, что:

1. Eliminator Boat Duel (USA) и Star Wars JVC (USA) зависают в зависимости от состояния выравнивания CPU/PPU на реальном железе.
   "Reading $2002 at the exact start of vblank clears the flag to 0 without reading back a 1. On most consoles and with most wait loops, an alignment is eventually reached such that the flag is read other than on at the exact start of vblank. However, Dendy-style PAL famiclones have a frame of exactly 113.667 by 312 = 35464 cycles, and 35464 is a multiple of 8. A bit/bpl loop that crosses a page boundary, such as that found in the game Eliminator Boat Duel, lasts 8 cycles. On some alignments, it hits the start of vblank every time and thus always fails to advance.
   So for the $2002 wait loop, do not make a wait loop whose length in cycles evenly divides the frame length."
   Эмуляторы, в которых имеется функция рандомизации выравнивания - в точности повторяют данный баг.

2. Ninja Ryukenden 1 (Japan) работает на железе с определенными артефактами и зависаниями в катсценах:
   https://www.youtube.com/watch?v=OdGUmJ4xr1M&t=86s - экран затемняется во время показа картинки меча в интро
   https://www.youtube.com/watch?v=OdGUmJ4xr1M&t=282s - зависон во время катсцены после окончания 1 уровня.
   При этом ни один эмулятор не может в точности повторить это поведение, зато вылезают такие баги:

• игра зависает при нажатии "start" в главном меню [Mesen 0.9.8, nintendulator, bizhawk, nestopiaue, fceux]
• эмулятор выдает illegal opcode после нажатия start, затем перезапускает титульник и уже работает нормально после второго нажатия start [Mesen 0.9.9, nintendulatorNRS, myNES]
• игра зависает сразу на старте выдавая illegal opcode [puNES, Mesen 0.9.9 (at some specific alignments)]
• Американская версия Ninja Gaiden (USA) НЕ имеет подобных проблем в эмуляторах.

3. Kyouryuu Sentai Zyuranger и Vice Project Doom/Gun Dec работают БЕЗ ГЛЮКОВ НА ЖЕЛЕЗЕ.
   При этом некоторые эмуляторы (NestopiaUE, My NES и Mesen (лишь при определенных значениях выравнивания))
   имеют проблемы с этими играми:

• Kyouryuu Sentai Zyuranger зависает совершенно рандомно между катсценами и переходами между дверьми в игре
• Vice Project Doom/Gun Dec зависает в конце 3-го уровня (в комнате с боссом)

Выяснить, что происходит в пунктах 2 и 3.
На основе этой информации будет возможно улучшить современные эмуляторы, которые имеют проблемы с этими играми. Разработчик Mesen писал:
"emulators are still pretty far from simulating stuff that occurs at a precise point within a cpu/ppu cycle, the alignment options in mesen don't do this either, even though it's definitely something that has impact on hardware"

ТАБЛИЦА:
excel-alignment.zip

As an example, implement some simple mapper as UCode to demonstrate and test the engine.

There are no sprites in the signal rendering.
Need to find out where they went and how to get them back.

Needed for operation without RegDump

Using the reference circuits: https://github.com/ogamespec/breaks/tree/master/Docs/NES

• Define ports
• CPU(APU) instance
• PPU instance
• Sim SRAM chips
• Sim aux. logic chips
• Bogus CIC

Cab you please add this? Here is a list of mappers to add:

• MMC1 (1)
• UxROM (2)
• CNROM (3)
• MMC3 (4)
• FME7 (69?)
• Maybe more later.
  Submappers can be easily seen on NESDEV WIKI,

Once everything is set up, add HLE for some of the circuits and increase FPS by several times.

Sometimes there are some white strips under the menu. Scrolling bug? SPR0 Hit?

Super Mario Bros. (JU) [!]_RegDump.zip

Reported by @eugene-s-nesdev

VRAM is constantly broken for some unknown reason by random values in random places.

(See "TIME LYFT" etc. in Bomberman)

Check.