Commodore used two different chips as the Commodore 64 CPU during its long lifetime. Both Commodore 64 CPUs were proprietary derivatives of the venerable MOS 6502, but unlike the 6502, Commodore and its MOS subsidiary never licensed either of them to other sources.
Like the Atari 2600 CPU, the 6510 and 8500 were 6502s with some changes. Unlike the 6507, the 6510/8500 added capability, rather than removing it.
The first Commodore 64 CPU: MOS 6510
The MOS 6510 dates to 1982-1988 or thereabouts. It was manufactured using the original MOS fabrication method, not to be confused with the name of the subsidiary who made it. It was a proprietary design and Commodore made them in its infamous MOS fabrication facility in Norristown, Pennsylvania.
The 6510 is a 6502 with added I/O to reduce costs through integration. Since Commodore owned its own chip plant, its engineers could tailor its chips to very specific needs.
Commodore never sold the 6510 to other companies, so Commodore’s competitors got by with the 6502 or other derivatives from MOS licensees.
You can identify and date a 6510 by its four-digit date code. The date code is the last line on the chip. The first two numbers are the week of manufacture. The last two numbers are the year. So a chip with a code of 2683 dates to the 26th week of 1983.
Some very early 6510s have ceramic packaging. Most use the common and familiar black plastic packaging.
The second Commodore 64 CPU: MOS/CSG 8500
The 8500 is a modernized 6510, manufactured using the slightly newer HMOS process. Adopting the HMOS process cut costs, but not enough to allow the Commodore 64 to sell for much less than $100 at any point in its life. Generally, the 8500-series chips used HMOS and 6500-series chips used the older MOS process. Other makes of 6502-series chips didn’t necessarily follow that convention.
The 8500 and 6510 are interchangeable. The 8500 runs a bit cooler than the older 6510 but otherwise is functionally identical.
Commodore Semiconductor Group, or CSG, is an alternative name for MOS. It was the same company, operating out of the same troubled fab. Commodore never got below a 3.5-micron manufacturing process, while Intel was making 8088 and 8086 chips with a 3-micron process during this time, and its more advanced CPUs on a smaller process. HMOS was on is way out at Intel around the time Commodore started bringing it in.
The 8500 was nominally the CPU in later Commodore 64C machines, although some late production 64s had older 6510s in them. Late in life, Commodore manufactured what it could with what it could scrounge up. If that meant outfitting machines with refurbished parts and old inventory, Commodore did it. Cashflow and continued existence was more important than consistency at that point.
Just like a 6510, the 8500 has a four-digit date code. The date code is the last line on the chip. The first two numbers are the week of manufacture. The last two numbers are the year. So a chip with a code of 5290 dates to the 52nd week of 1990.
Background on MOS Technology
Commodore bought MOS Technology in 1976 to get a chip supplier. Commodore used Texas Instruments as a supplier but found itself competing against TI. Buying MOS allowed Commodore to achieve the same kind of vertical integration, and put it in the enviable position of selling chips to competitors like Apple and Atari.
Owing its own chip design house and fabrication plant gave Commodore a competitive advantage in the early 1980s. Owning MOS was the only reason Commodore was able to price the Commodore 64 at $595 in 1982 and cut prices to half that by the end of 1983. But as the decade wore on, Commodore couldn’t necessarily make chips cheaper than it could buy them anymore. By 1986, it was questionable whether owning MOS (also known as Commodore Semiconductor Group) still provided any advantage. Commodore closed MOS/CSG in 1992. Post-1992 Commodore 64s were made from old inventory.