The Western Design Center 65c816 is an underdog CPU from the 1980s. It was never the best available CPU of its time and it was never the cheapest. It was a 16-bit CPU from a time of transition from 8 bits to 16 and 32 bits, released around the same time as the first fully 32-bit CPUs. But it’s an interesting CPU, even if it doesn’t get the attention other contemporary CPUs received. It was introduced the week of December 12, 1983.
The 16-bit 6502
The 65c816 is interesting because it’s an extension of the MOS 6502 to a full 16-bit architecture. The key designers of the 6502 stayed at MOS Technology after Commodore purchased the company, but starting in 1980 they realized they had better opportunities elsewhere. Chuck Peddle left MOS and Commodore in 1980. Bill Mensch left in 1981 to found his own company, Western Design Center. Mensch’s projects included the 65C02, a CMOS version of the 6502 that could reach higher clock rates while using less power, and the 65816.
The 65816 combined the 6502’s efficiency with the ability to behave like either an 8-bit or 16-bit CPU. It powered up in compatibility mode, where it acted more or less like a 65C02. Well-behaved software that didn’t use the 6502’s undocumented opcodes generally ran just fine on a 65816. Truth be told, programs that didn’t run on a 65816 didn’t run on a 65c02 either.
In native mode, it had a 24-bit address space, allowing it to address up to 16 megabytes of memory. This was a huge amount in the mid 1980s. More practically speaking, it meant you could put more than 64K of RAM in a 65c816-based system and not have to do any special tricks to address it. A 6502 could only see 64K at a time. So if you put more than 64K in it, it had to divide it up. The 65c816 could see and use its full memory space all at once, whether it was 256K, 512K, or several megabytes.
The 65c802 variant of the 65c816
Western Design Center also offered a 65c802 processor, which was a cut-down 65c816. It had the exact same pinout as a 6502, so you could replace a 6502 with it. It had all of the instructions of a 65c816 and could operate as either a 6502 or as a full 16-bit 65816. But it only had enough pins to address 64K of RAM.
You could upgrade a 6502 or 65c02-based computer a little by installing a 65c802 in it. Existing programs wouldn’t run any differently with it, but if you wrote your own software, you could take advantage of the 105 additional opcodes the 65816 instruction set offered to do complex operations faster.
Advantages and disadvantages of the 65c816 CPU
The 65c816 CPU had some advantages and disadvantages compared to other CPUs of the time. Clock for clock, it was very efficient, so a 65c816 running at 8 MHz was faster than a 68000 or 80286 CPU running at the same speed. And since all three chips had a 24-bit address space, they all had the same memory limitations.
Being based on the 6502 was a boon too, if you were already familiar with programming the 6502. Some developers hated the 6502, but if it was what you were used to, the 65c816 was a pretty easy transition. Backward compatibility with the 6502 was useful if you had existing designs based on the 6502.
But it was an evolutionary dead-end. While 32-bit versions of the x86 and 68000 family appeared around the same time as the 65c816, the 32-bit version of the 65c816 never went into large scale production. It’s been reverse-engineered and exists as a core you can use today in an FPGA, but it wasn’t available for use in the 1980s and 1990s.
It also appeared a bit late. A common question is why Commodore didn’t use the 65c816 in the Commodore 128. There’s a simple reason for that. It didn’t ship in quantity until late 1985, and Commodore needed to get the C-128 onto the market in early 1985 so they’d have something to sell until the Amiga was ready. If you’re wondering why IBM or some other company didn’t use it, it was probably for a similar reason.
Who used the 65c816
The two most successful products to use the 65c816 CPU were the Apple IIgs and Nintendo Super NES. Apple used the chip as-is, while Nintendo used its core in a custom chip. In addition, WDC sent samples to Atari at the same time they sent it to Apple. Presumably Atari considered using it in some sort of a souped-up version of the Atari 800. Atari didn’t use it, probably because around the time it received the samples, Jack Tramiel bought the company, and Tramiel wanted to sell a 68000-based computer.
Commodore considered using the 65c816 in a computer called the Commodore 65. But the design was a bit late and Commodore never shipped it in quantity. It sold the prototypes in late 1993, when it was in dire financial condition and sold pretty much anything it could find to try to stay in business one more quarter.
Holding it back
Both the Apple IIgs and Nintendo Super NES were interesting machines and the SNES in particular was very successful. Both machines were less than they could have been, running the CPU at 2.8 MHz or 3.58 MHz. Running a 4 MHz CPU at 3.58 MHz is a defensible position, since it’s exactly double the rate of a 1.79 MHz clock they needed anyway. But running the chip at 2.8 MHz made the Apple IIgs an underachiever. Apple didn’t want it to be too attractive compared to a Mac. Running it at 2.8 MHz allowed it to be faster than an Apple IIe, but less than half the speed of a Mac.
Upgrade products based on faster versions of the 65c816 existed for the Apple IIgs and even 6502-based systems like the earlier Apple IIs and the Commodore 64. They were too expensive to be mass market products, but they allowed enthusiasts to extend the useful life of their favorite systems. The upgrade products showed what the IIgs could have been if it Apple hadn’t downclocked the CPU.
Use of the 65c816 CPU today
The 65c816 is still in production today, for use as a microcontroller or for other industrial or embedded applications. Its days as a CPU for mainstream computing are long gone, but it’s a familiar chip with good development tools and it’s very inexpensive, making it ideal for those specialized uses.
David Farquhar is a computer security professional, entrepreneur, and author. He has written professionally about computers since 1991, so he was writing about retro computers when they were still new. He has been working in IT professionally since 1994 and has specialized in vulnerability management since 2013. He holds Security+ and CISSP certifications. Today he blogs five times a week, mostly about retro computers and retro gaming covering the time period from 1975 to 2000.
reviews of Apple IIgs complaint with how slow 65c816 CPU was working
Bill Mensch approached Commodore about using the 65816 in the C128 but was firmly rebuffed by Bill Herd and Fred Bowen. Later on, Dave Haynie was hoping to use it in the C256, a successor to the C128 which ended up being cancelled.