Inside the Famicom | 02: The 6502 CPU

[@play_history]

Okay so on the BCD removal thing: To my understanding this was a redesign that Rockwell did first. MOS second-sourced production of the 6502 and variants to Synertek and Rockwell (a common practice in the industry at the time). Rockwell made modifications to the chip and Rockwell sub-licensed the rights of their design to Ricoh. MOS engineers have claimed that Nintendo stole the 6502 because they weren't getting direct payments, but anything they would have gotten would have gone through a long chain through Rockwell. Such are the salad days of early microprocessors.

[@whatskenmaking]

Yeah, the re-licensing of the tech makes it really difficult to pin down what exactly happened. I read through a couple of accounts and Bagnall’s seems to at least summarize the reaction from Commodore employees with a firsthand account

[@AlexanderEmashev]

Wow, such a thorough and clean explanation of how the CPU works. Animation really helps to understand what happens and why. Thank you! Looking forward to watching the next part.

[@whatskenmaking]

The animations take a lot of time, so I’m glad they’re useful!

[@SplitHouse]

This is a great video to watch while I'm working on my NES emulator project! Thanks!

[@HerrBerzerk]

Really great explanation. Your didactic skills are fantastic. I need to learn more about how these old CPU-Board works in detail.

[@scotchnichols]

I've really enjoyed your deep dives into the Famicom so far. Looking forward to the next one.

[@HSTVizle]

These videos are awesome! My favourite lecture was Microprocessors in college, we've studied similar 8-bit CPU desings. Keep going Ken!

[@rudy5360]

Loving this series! Watching along as I plan to AV mod my newly acquired Famicom!

[@ZipplyZane]

What I love is that the lack of hardware BCD is why Tetris eventually crashes. It's the software BCD that winds up eventually taking too long to calculate and results in corruption. So that whole Tetris crash thing made something this old relevant again. Also, I love that it's the same chip with that line cut, and that was enough back then to say it was a different chip. That's the only reason it was left out. It allowed the copy to be legal.

[@willllbert1]

Fantastic series! Looking forward to learn more 😊

[@TheMikeyb86]

Love this. Keep 'em coming. Thank you.

[@leandrormor]

Thanks for the content. If the CPU divides the master clock input, does that affect the output? I'm curious to learn what clock reference the others components expect since cpu internally uses a different clock.

[@kauegoncalves9532]

fantastic video !

[@omarhossam9199]

This is so goated omg. What a great video

[@wizzo-b8i]

Really enjoyed this! Not a huge fan of the games on these consoles but I have a huge soft spot for the 6502

[@Twenty_Six_Hundred]

Great stuff!

[@arnauddemarais9104]

Great great great video. Thanks !!!

[@chansonjoy]

Interesting, in fact it's very close to modern MCU architecture, except very limited IRQ resource and probably doesn't support nested IRQ at all. Modern MCU usually also come with NMI IRQ, but I found it hardly being used, because there are many other high priority IRQ available. I wonder what's the scenario of NMI in 6502.

[@whatskenmaking]

You can nest IRQs if you reset the interrupt flag in your IRQ processing. For the FC and NES, NMIs are used by the PPU for Vblank signaling. I’ll talk about that when I cover the display subsystem

[@iwanttocomplain]

There's something I've been trying to remember about 6502 bus requests for a while and not into all the resarch and it's really the key issue I have with the 6502. The SNES has 128KB of working ram. People think it's all available. But it's not. I'm also quite proud of myself for predicting two 8KB buffers in the NES by counting the numbers and not figuring out how it could have 2K of work and video and also 8K, but slower.

[@whatskenmaking]

I'm going to walk through how the address decoding works in the next video - that will explain why there's an 8K chunk of memory with only a single 2KB RAM chip in it 😉

[@iwanttocomplain]

@@whatskenmaking I'm speaking of the 6502 bus access system

[@whatskenmaking]

Yeah - the address decoder logic is why the bus behaves that way. In short, chip select is enabled on work RAM through a higher order bit, but some of the lower order address pins aren't connected. I'll have some visuals that walk through it

[@iwanttocomplain]

@@whatskenmaking yes that's it. I really want to understand why the snes has 128KB ram.

[@skilz8098]

If I remember correctly, don't they use a Zero Paging Scheme? Something on the lines like when the system writes information to Page 0 based on the chunk or page size, it also writes to all other sections as well. It's been quite a while since I've done this type of research and read into, so please feel free to correct me if I'm wrong. But I think I remember learning something about this with their memory mapping scheme. This is kind of what tripped me up the most within the 6502 Assembly for the NES where I've been more familiar with and accustomed to x86 ASM (Intel Syntax).

[@cosmefulanito5933]

Since the Z-80 was and still is today the most used processor in the world (enbedded today), I don't understand why people in the USA are so fixated on the 6502. Good video.

[@whatskenmaking]

I don't know - maybe because of the popularity of Commodore or the Apple II here in the US? Variants of it were used by Atari as well. Personally, I've always been more of an Intel guy

[@ScottDuensing]

Simple! It was cheaper!

[@skilz8098]

One of the things I think is driving it isn't just due to the various devices that used it throughout the 70s and 80s almost up into the early 90s, but also the fact that it does have a huge part in the History of Computer Science - Electonics Engineering. On top of that, there's a lot of information on it, it is well researched and well documented. There's also the sense of nostalgia that comes from it. Finally, one of the most intriguing aspects is that the original 6502 was physically designed by hand. Every single transistor and wire were laid out by hand. They didn't use a sophisticated piece of software to automate this process. I think this is why it is so intriguing. Sure, the Z80 is just as intriguing but so is the 8080 by Intel. They are all feats of Engineering! They are all worth learning about.

[@bloepje]

As a programmer I would chose a 6502 above a z80 anytime. And I did a lot of z80/8085 and 6502 reversed engineering. I thought the 8052 was one of the most used cpus for embedded applications. I mean that stuff is embedded in switches as either the controller of a dumb switch or the interface between the switch and the cpu of a managed switch.

[@bloepje]

To be clear, I did RE on 8052 too... that was a deep dive in a controller where banking is embedded in the ISA

[@stempek121]

What's the software shown at 2:26 mark?

[@whatskenmaking]

That’s Visual2A03 - there’s a link in the description. You can give it a set of instructions and it will show you what’s happening on the chip as they’re being executed.

[@popcorn109i]

Epic

[@docwhogr]

the nes inner working was explained by many, i think someone should give love to the other consoles. you did an excellent job though when you said go make something cool i went and fill the icecube tray in my fridge. does it count?

[@whatskenmaking]

Thanks! I do plan on doing a series like this with other consoles and computers, as well. I started with the Famicom series for a couple of reasons, but mostly because the NES was my first childhood game console. Putting your ice cube tray in the fridge won't do much - you might try the freezer next time; that would be cooler 😆

[@ssl3546]

Chip masks are only protected for 10 years and the Famicom came out before the mask law was even passed. It would be foolish not to use the masks or to pay a license for something in the public domain. Do you send KZbin money for the free music in your videos?

[@JamesSjaalman]

IMHO Great Britain did not have the PAL standard.

[@mwk1]

So... if the Famicom 2A03 chip has 16 bits wide address bus does it mean that it really is a 16-bit CPU? (hello Atari Jaguar!) 😎

[@whatskenmaking]

The data bus is what makes it an 8-bit system. The 16-bit address bus just means it can address 64K of memory… but it only does it 8 bits at a time.

[@eitantal726]

@@whatskenmaking I thought it was register size

[@bloepje]

The jaguar was 68k based I think and that is a 16/32 bit cpu. It had a 16 bits data bus but was mostly 32 bits in calculating. Somewhere in history the definition changed from the size of the external databus to the size of the common alu size. The size of the external databus became much bigger than the alu size on amd64 ISA. Yet we don't call the 128 bits or larger. The 8088 had an 8 but databus but was internally an 8086, so a full 16 bit cpu. At that time we were not sure if we should call it 8 bits or 16 bits. We called it 8 bits if you don't like the ISA and we called it a cheap 16 if you liked it. Yes there were actual camps of this cpu is better. Intel made 4 interesting cpus: the 8089, the 80860, maybe the 8096 and the 80960. The 8088/8086 was so riddled with z80 shit it wasn't fun. There were a lot of interesting ISA's and the worst became popular. I know that back in the late 80's, early 90's we tried to get access to the T800 transputer. Or actually any. Even by today's standards, they were very innovative.

[@truectl]

This isn't "begging the question."

[@whatskenmaking]

Perhaps 😊

[@jeff.fredrickson]

Current vernacular uses “begging the question” to mean “raises the question” even if that meaning has nothing to do with the original meaning. Languages evolve even if the evolution was based on improper usage.

[@TheMysteryDriver]

To beg the question and begging the question are now two separate things.