VerilogBoy - GameBoy on FPGA

This project consists of two parts:

An open source Game Boy® compatible console Verilog RTL implementation (the VerilogBoy Core)
An open source FPGA-based handheld game console hardware (the VerilogBoy Handheld, or vbh for short.)

Both are documented in this project.

Original Goals for the CMPEN 275 course project

This project aims to recreate the whole Game Boy gaming system on an FPGA development board, with the ability to play commercial game like The Legend of Zelda with no major glitches.

To be specific, it should be able to run the unmodified Game Boy machine code, produce gray-scale graphics and output to an external monitor, produce the sound and output to the 3.5mm jack on the FPGA board, and accept user input to control the game. Other functionalities like serial communication and IR communication are currently not part of this project.

Current Goals

Besides getting it running on a development board, I am designing a dedicated handheld console for this. The console is simply called, the VerilogBoy Handheld. GameBoy Color support and serial communication capbility is also in the long term plan.

System Architecture

The main system architecture is designed as follows (outdated):

There are three major parts needs to be implemented: the Game Boy CPU (8-bit CISC Processor, Intel 8080 like), the PPU (or GPU), the Sound unit. Several interfacing modules are needed to support the IO capability provided by the FPGA development board. Game ROM would be stored in on-board NOR flash, and RAM would be implemented with on-chip Block RAM.

Specification of the Handheld

XC6SLX16-FTG256 FPGA
8MB ISSI IS66WVE4M16 PSRAM
Wolfson WM8960G Audio Codec
XPowers AXP209 Integrated Power Management Unit

Progress

Able to run Is That a Demo in Your Pocket with sound. See demo video! (Please turn down the volume, I found the signal from the FPGA is too hot for my recording device).

Pokemon Yellow and The Legend of Zelda: Link's Awakening DX are also tested to work with DualShock 2 support ( I should have picked a Nintendo(R) controller instead of Sony(R) one... But that's what I have. ).

After finishing the presentation, I started refactoring the code, starting from the CPU. I also started working on the handheld as described before. They are all still working in progress as of the time I am writing this. You can follow the progress on this page and on the GitHub commit history.

Project Logs

Collapse logs

Demo firmware for Pano Logic G1
Wenting Zhang • 04/14/2019 at 17:11 • 1 comment
Upcoming demos and talk
Wenting Zhang • 04/10/2019 at 00:04 • 0 comments

Hi all,
It has been another 2 weeks since my last update. A lot of things are going on for this project:
The USB host stack is working, with working USB HID driver and USB Mass Storage Driver
The refactored VerilogBoy CPU is working now, passing all unit tests from the first revision, plus Blargg's CPU test.
These two together, means there are some games actually running on the machine:
I am planning to do a writeup about the USB, and in the future maybe unit test, but currently, most important:
I will be doing a demo of VerilogBoy at VCF SE during Apr 27-28 and LatchUp during May 4-5.
I am also going to give a talk about this project at LatchUp.
So now the top priority for me is to get these things right, means I would probably not have time to write something here. Anyway, thanks for reading this quick update, and hope to see you guys at VCF SE and LatchUp!
A detour to Pano Logic G1 (3) - UART & Hard fault
Wenting Zhang • 03/26/2019 at 00:40 • 0 comments
It has been a while since my last update. I have been working on the USB stuff for Pano Logic G1, mainly for connecting to joysticks and flash drives. I was concerned that my LPDDR and cache would cause me some trouble when the code is being executed in RAM, but so far they are holding up well. I will talk more about them in the next log. In this log I would like to talk a little about some debugging utilites, namely, the UART and the hard fault.

UART

UART is very handy when you want to see logs from the device. At first I thought I can get away just by using VGA text terminal, but it soon turns out that 80x30 text is simply not enough. Unfortunately the Pano Logic doesn't have any serial ports. From the schematics, it seems that they originally have one, but was removed after some revisions. But anyway, I have to repurpose the IOs to create myself a serial port to use.

This is not new to Pano Logic, Skip Hansen from PanoMan project has already done this: he soldered a wire to the LED pin and get the serial output from there. For me, as mentioned in one previous log, I do not have soldering iron with me currently, so I need to find some other way.

As an alternate, I used the wire clip come with my logic analyzer. They can be attached to through-hole components easily, such as this VGA connector.

I am using VGA SCL pin for the serial port. I wrote an extremely simple UART transmitter to transfer the data: https://github.com/zephray/VerilogBoy/blob/refactor/target/panog1/fpga/simple_uart.v. Why I don't just use Skip's UART transmitter core for Pano Logic G1? Or why I don't just work on top of Tom and Skip's project? Well, the (stupid) answer is the same regarding why I picked PicoRV32 rather than VexRiscv: I have decided (long ago) to call this project VerilogBoy, so all the source code should be written in Verilog, not SpinalHDL. Yes I am also aware there are tons of better open-source UART controller written in Verilog available online. I am probably just too lazy to find one considering I don't need other fancy functions anyway.

Using this UART transmitter also very easy, hook it up to the bus and write to the only register available: data register. It has too possible operation modes depending on the way it is connected. If the ready signal is connected to the ready signal of the bus, the UART transmitter would block the code execution until it finishes the transmission. Means the UART print function can be as easy as:
```
#define UART_TXDR *(volatile uint32_t *)0x03000100
void uart_print(char *str) {
    while (*str) UART_TXDR = *str++;
}
```
There is no need to worry about if the transmission has finished or FIFO overrun whatever, since it simply throttle the write speed to transmission speed.

Alternatively, if the ready signal is connected to the external interrupt input of the CPU, it can generate a end of transfer interrupt, then a software FIFO can be implemented.

Hard fault

Are you familiar with the Hard Fault in microcontrollers, or Segmentation Fault in Linux, or "The program has stopped working" in Windows? Usually they are referring to the same thing: the code is accessing some memory they shouldn't touch. Most commonly, dereferencing a null pointer (which is nothing more than a pointer points to address 0). I hate this error: they are so common, but not easy to debug. However, I have to admit, it is the one who points out the bug in my code, debugging would be even harder without anyone telling me where goes wrong.
(It simply hangs when such things happens. It can be hard to determine who caused the issue on real hardware.)

Unfortunately, the processor core itself cannot detect such errors, as it doesn't really know what is right or wrong. Generally, MMU or MPU is in charge of this: the program defines the valid address range (physical address or virtual address, depending on the specific environment.). The MMU or MPU would then generate an exception when an illegal memory access happens. On systems without...
Read more »
A detour to the Pano Logic G1 (2) - Cache
Wenting Zhang • 03/02/2019 at 16:48 • 0 comments
As one of the conclusion of the last log, in order to use the LPDDR memory in 32-bit mode on the Pano Logic G1, a cache is almost a must. Sure I can just use 16-bit mode, half the capacity (16 MB) isn't really an issue for me... But I still decided to just implement a cache, it shouldn't be that hard.

So, as a result, I have got cache working on Pano Logic G1. It is a 8-KBytes 2-way set-associative cache. Replacement policy is LRU and write policy is write back. (The whole point of having a cache is because write through is almost impossible given the data mask cannot be used.) It is connected between the PicoRV32 CPU and the MIG memory controller, so all read and writes to the LPDDR is cached. I won't go into details about the cache since I feel like there is nothing special worth talking about except being slow and inefficient. I will add an bus master arbiter between the PicoRV32 and the cache in the future, so the GameBoy CPU could access the LPDDR as well. Though one need to keep in mind this is only a 2-way set-associative cache, having multiple masters would lead to very questionable performance.

So, what about the performance? Currently:
- Read hit: 2 cycles
- Write hit: 2 cycles
- Read miss: 4 cycles + memory read latency
- Write miss: 5 cycles + memory read latency
- Read miss + flush: 12 cycles + memory write latency + memory read latency
- Write miss + flush: 13 cycles + memory write latency + memory read latency
So you can see.. The cost of missing is high, and the cost of flush is very high.

Also, due to my bad coding and the limitation of Spartan-3E's block RAM (it does not support byte enable, which is important for a cache that allows byte enable), compared to 16-bit non-cached version, the whole design uses 1500 more LUTs. I assume mostly comes from the cache, and some from the 32-bit memory controller.

But any way... It Works™.

How to use:

Make sure g++, riscv32-unknown-elf-gcc, and ISE 14.7 are installed. The RV32 gcc should support march=rv32i.

Clone the VerilogBoy GitHub repo, check out commit b08377d (Merge branch 'refactor').

Run the following command:
```
cd tools/bin2mif
g++ -o bin2mif bin2mif.cpp
cd ../../target/panog1/fw
make
cp *.mif ../fpga
```
Go to target/panog1/fpga, open the project pano_top.xise with ISE 14.7 and generate programming file.
A detour to the Pano Logic G1 (1) - LPDDR
Wenting Zhang • 02/24/2019 at 16:42 • 9 comments
As I have mentioned in the previous update, I am still having some trouble with the MIPI-DSI. Currently I do not have access to any soldering tools, so the plan of making a new revision of prototype need to be postponed. In the meanwhile, I thought it might be a good idea to continue working on the RTL – I started refactoring the code but haven’t finished yet. But, I need a hardware platform to test. Well, I forgot to bring my FPGA development board (Xilinx ML505, I really loved that board) with me when I came back from Christmas holiday… But no problem, I got myself two Pano Logic thin clients (G1 and G2) last year. Though I have to admit, I didn’t do much with these units after I got them. Now the time has come, let’s take a look.

We have something to hack

(Image showing is my own Pano Logic G1)

In case you are not familiar with them, let me introduce them first. They were originally thin clients, used to connect to remote desktop servers. What is special about Pano boxes are, they are powered by FPGAs, rather than ARM or x86 CPUs commonly found on a thin client. They advertise it being a “Zero Client”, means there is no (zero) software running on the client. Well, unfortunate for them, they went bankruptcy in 2013. What is fortunate for us is that, these units now become useless for companies originally bought them, being sold for very low price on places like eBay. It is our turn to repurpose these devices! Of course, hackaday has already featured it for several times: https://hackaday.com/2013/01/11/ask-hackaday-we-might-have-some-fpgas-to-hack/, https://hackaday.com/2018/12/07/racing-the-beam-on-a-thin-client-in-fpgas/, https://hackaday.com/2019/01/11/pac-man-fever-comes-to-the-pano-logic-fpga/, and https://hackaday.com/2019/02/11/two-joysticks-talk-to-fpga-arcade-game-over-a-vga-cable/.

As far as I know, there are 3 generations of Pano Logic clients, the first two looks very similar, and the third is slimmer. Unfortunately I have never seen a slim model on the eBay. If you know anything about the slim model, please tell me, I am interested. The first generation (G1) model is powered by a Xilinx Spartan-3E XC3S1600E FPGA (1600K system gates, translate to around 30K LUT4s.), with 32MB of on-board LPDDR RAM. The second generation model, depending on the revision, is either powered by a Xilinx Spartan-6 XC6SLX150 (Rev. B) or Xilinx Spartan-6 XC6SLX100 (Rev. C), both with 128MB of DDR3 memory. The one I own is a Rev. C one. Both generations has already been reversed engineered by the community, notably cyrozap, twj42, and Tom Verbeure. You may find more information about details of the Pano boxes here: https://github.com/tomverbeure/panologic, and https://github.com/tomverbeure/panologic-g2.

Now, which generation should I focus on? Gen 2 is significantly more powerful, however getting harder to find on eBay. The Gen 1, powerful enough for my purpose, and can still be purchased easily on eBay. I decided I want more people being able to play with my VerilogBoy code (if any), so I will go with Gen 1. Also, all the framework I developed for Gen 1 devices might help others looking into playing around with their own G1s.

Talking about the G1...

Then, the gen 1 has already been reversed engineered, someone even published its schematics online, it should be trivial to just port the existing code to the G1, right? No. There are still several issues to be solved:
- The G1 does not have any GPIO for user. In order to attach a game controller, one would need to repurpose some of the IOs (like in the panoman project, he used the I2C from VGA port), or use a USB joystick. Which means I need a host-side USB stack running on a soft core on the FPGA. As I have said, I hope more people can play this, so I will go with the USB solution.
- The G1 doesn’t have any on-board storage large enough to hold...
Read more »
Revised handheld hardware architecture
Wenting Zhang • 02/02/2019 at 14:36 • 3 comments
This is a project update about the hardware side of the VerilogBoy Handheld.
After testing the previous prototype (Rev 0.1), I feel like several changes are required:
- Add a dedicated DPI-to-DSI bridge chip. My poor implementation of the D-PHY transcevier simply couldn't meet the signal integrity requirement. This is an experimental change to test how much it can improve without doing major change to the board (for example, moving to 6-layer, adding decoupling capacitors that would interfere with overall structure, etc.).
- Replace the Micro-Type-B USB socket with Type-C socket. No Alt Fn or PD support is planned, just working under USB 2.0 FS slave (sink) mode.
- Add a microcontroller to handle hardware initialization, RTC, and USB FS communication (for example, flashing new firmware to the on-board SPI flash.
- Fix various incorrect component footprint.
Here shows a revised hardware architecture. This is probably overly complicated for a hobby project.
Due to the closure of PCB manufacturers because of the Chinese New Year, we are probably not going to see the new prototype (Rev 0.2) in the Feburary. I will continue working on the HDL side of the prototype and trying to finish the CPU refactoring within the coming weeks.
For the time being, thanks for reading.
Hello from the DSI screen
Wenting Zhang • 01/28/2019 at 17:41 • 0 comments

Here it is. 320x320 IPS MIPI DSI screen. DSI running at 268 MHz (256 MiHz), that's exactly 64 times the Game Boy pixel clock rate (4 MiHz). With every GameBoy pixel quadrupled, and every pixel being 16 bits (FYI, the GBC uses 15bpp), it is going to transmit the pixel output from the PPU to the screen perfectly in sync. DSI controller is implemented in FPGA. A custom boot ROM running on the GameBoy CPU takes care the DSI controller initialization as well as screen initialization. More details about DSI is coming.
The Assembled PCB.
Wenting Zhang • 01/14/2019 at 02:11 • 1 comment

Unfortunately due to the system design, I have to finish refactoring the CPU first before I can do some practical tests (which all need the BootROM to run on the CPU within the FPGA). But, at least, here is a nice looking board!
On the way designing the hardware
Wenting Zhang • 10/22/2018 at 23:40 • 0 comments

Finally I decided to design the hardware for it, rather than just running on the devboard. Here is the short video provides a time lapse of me designing the hardware, and a short introduction to the project.
DualShock controller working
Wenting Zhang • 04/16/2018 at 03:39 • 0 comments

For the controller, I decided to use the PSX controller (DualShock 1). I know that I could use something like a NES controller, which is probably easier to interface with (?), but I just thought that it might be used in my future projects, having more buttons is probably better. So, as I result, I got a cheap DualShock 1 clone off eBay.
I spent some time writing and testing the driver for it, and now it is working (partially). I am able to read all buttons out, including two analog joystick. However, I am unable to enter the escape mode and manually set it to work under analog mode rather than digital. The controller simply refuse to ACK my 0x44 Set Major Mode command. I guess this is not a big deal, the Game Boy does not even need the analog joystick, and I can always set the mode by pressing ANALOG key.
Here is a screenshot of the oscilloscope showing it working in the analog mode:
And here is a screenshot of it refuse to ACK my 0x44 command:
For more information about the driver itself, please refer to the code.