4 ms·

I don't wish to detract from Chuck Peddle's work, because the 6502 really is an amazingly efficient and elegant design. But I've always felt the microprocessor

by retrac 7y ago

I don't wish to detract from Chuck Peddle's work, because the 6502 really is an amazingly efficient and elegant design. But I've always felt the microprocessor has been overstated in its importance in the computer revolution. Throughout the 60s, 70s and 80s, memory was usually considerably more expensive than the CPU, for all kinds of computer system. The majority cost in mainframes, minis and embedded machines alike was the core memory. Early semiconductor RAM was even more expensive than core! It was possible to design a simple CPU in the early 70s using a few dozen SSI/MSI chips, and a board with such a CPU could have been done in less than 30x30 cm. Indeed, that's basically a description of the numerous low-end MSI-based minicomputers of the 70s. The cost of such a design would have been in the same ballpark as many of the early single-chip microprocessors released in '74 and '75 before the 6502. But what use would such a processor be, whether discrete or a single-chip microprocessor, without memory? A mere 4 kilobytes of DRAM in 1973 would have cost over $1000, and a useful general-purpose computer really needs more like 8 or 16 KB to host an assembler and disk-operating system. By late 1978, the cost of 4 KB of semiconductor DRAM fell to $60. By 1983 a mere $2. In many ways, that was the true revolutionary technology. Even if no one had thought of microprocessors, and even if advances in CPU speeds had petered out, a minicomputer with 2 MB of RAM is an entirely different kind of beast from one with only 4 or 8 kilobytes.

walrus01 7y ago
For comparison of costs and dollar figures, $1000 was a huge amount in 1973. I recently did the inflation math on a 1978 advertisement for a $699 TRS-80 bundle (with cassette tape drive!), adjusted for today it would be equivalent to almost $2800, enough to build a really powerful gaming desktop.
jacquesm 7y ago
The 6502 was an enabler if there ever was one. Yes, memory was expensive. But we used only a very small fraction of what we think of as normal today. A typical system would have 2K ram; maybe 4K and if you were rich you'd splurge for the 16K upgrade (but what would you do with so much memory?). In comparision, the 6800 sold for a large multiple of the 6502, and Motorola wouldn't have slashed their prices if not for the 6502. Peddle & Co showed that a CPU did not have to be expensive and that was a game changer. A whole generation of 8 bit machines depended on it: VIC-20, CBM 64, Atari, Acorn Atom, BBC Micro, Apple I, Apple II. None of those would have happened without an affordable CPU to power them. Very few chips can claim such a distinguished legacy, and without showing the market potential for this 'personal computer' thing IBM might not have entered the game at all. Visicalc (which required 32K!) was the breakthrough moment.
- agumonkey 7y ago
  It's not hard to see the effect on ram price when you have all these tiny personal computers broadening the market. Synergy I say.
whyenot 7y ago
But we also didn't need as much memory. The memory resident portion of CP/M was only 2kb, and that is with some inefficiency because it was necessary to stick with the 8080 instruction set. Just recently there was an article on HN about the passing of Randy Suess, who along with Ward Christensen created the first real BBS. The other part that often isn't recognized, is the work that Irv Hoff did to re-write the memory resident part of CP/M, so that it not only did everything that a operating system needs to do, but could also accept remote connections, deal with a modem, and have some (very rudimentary) security. All in 2kb. That's amazing.
- retrac 7y ago
  > But we also didn't need as much memory. Another way to look at it is that we only did the things we could do with such little memory. CP/M is a marvel of size-constrained code, but it's also crude in the extreme. A multitasking virtual memory system with a decent file system and an optimizing high-level language compiler -- what most developers would consider the bare minimum for being civilized today -- really relies on having about a megabyte or more of RAM in practice. Consider the web, one of the "killer apps" for computing in general. It's not hard to imagine implementing something like modern packet-switched networking and basic text-only hypertext documents, with a computer which can only do about 100,000 instructions per second. But doing it with only 4 or 16 KB of RAM, though? Just this comment is 843 bytes.
  - whyenot 7y ago
    MP/M didn't have virtual memory, but it had everything else on your list, was released in 1979, and had a minimum memory requirement of 32kb. In the mid 1980s, I ran a four line BBS on an Atlas ACS8000-14 running MP/M. I wrote the BBS in DR CBasic. I attempted to port it to Turbo Modela-2, another high level language, but the IDE was so buggy that I eventually gave up (being female and a teenager, there was also a lot of social pressure not do computer stuff).
  - wbl 7y ago
    The PDP-11 had 16k.
tyingq 7y ago
The Z80 was around $65 when the $25 6502 came out.
- jacquesm 7y ago
  And the 6800 was $175, and with the $ exchange rate of the time it was unaffordable here.
pvg 7y ago
such a CPU could have been done in less than 30x30 cm. That's far too big for just the CPU of a home computer. Look at the form factors of the C-64, the Apple ][, the TRS-80, the ZX Spectrum, etc. You had to be able to take it home and plug it into your TV. Something the size of a mini fridge is not really a 'personal' device.
- aidenn0 7y ago
  30x30 would fit comfortably in those cases, so put it on a riser card and only suffer an extra couple of inches of thickness. Of course the manufacturing price would suffer. I don't see someone making an LSI populated PCB for $25
  - pvg 7y ago
    Most of these don't have two sides that are 30cm. The only reason the Apple ][ does is the expansions slots.
    - vidarh 7y ago
      The price of several of them also very much depended on the low cost of assembling small, simple boards. Commodores price points would have been impossible with larger boards with more components, for example. The Apple II was crazy expensive in comparison to many of the other 6502 based computers.
AnimalMuppet 7y ago
Yes, you could build a CPU from a number of MSI chips. But what kind of clock frequency would you get from that kind of a setup? One chip gave you much better performance. A multi-chip CPU would be a toy; a one-chip CPU could do real work (by the standards of the time).
- retrac 7y ago
  > But what kind of clock frequency would you get from that kind of a setup? One chip gave you much better performance. You might be surprised, because the answer is in the ballpark of 100 MHz when cost is no object. When cost is an object, well even simple designs like the MSI/LSI versions of the PDP-8 were clocked around a 1 microsecond memory cycle, comparable to most early microprocessors. For an extreme example, the Cray-1 was constructed entirely from tens of thousands of discrete NOR gate chips using ECL, and was clocked at 80 MHz. Multi-chip implementations remained standard on big systems into the 1980s. The ability to use bipolar technology over the much slower (at the time) MOS was a major factor, but it also simply wasn't possible to add to a single die sophisticated features like large register sets, pipelining, caches, etc. with the limited transistor budget available in the 1970s. It wasn't until the mid-80s that CMOS microprocessors started to win decisively versus discrete logic for mid-range applications. It lasted even longer on the very high end -- DEC, IBM, Cray, and others, were manufacturing processors constructed out of many MSI/LSI chips well into the 1990s. Even the first POWER processor from IBM consisted of 9 VLSI chips, in 1990. The overwhelming advantage for single-chip processors back then was the low cost. A minicomputer like the PDP-11/20 from all the way back in 1970 constructed out of SSI chips could run circles around the 6502, but the processor for the PDP-11/20 was like $2000 and the 6502 was $25.
  - Taniwha 7y ago
    I completely agree with all your points, but would add that we also didn't have the CAD tools to make big stuff, the 6502 was built by cutting litho to make transistors (actually physically cutting different coloured transparent tape and sticking it to a transparent sheet), nowadays I could knock out all fully timed 6502 in a few days with the right tools. Building something the size of a Cray on paper was an amazing achievement

by retrac 7y ago

walrus01 7y ago

For comparison of costs and dollar figures, $1000 was a huge amount in 1973. I recently did the inflation math on a 1978 advertisement for a $699 TRS-80 bundle (with cassette tape drive!), adjusted for today it would be equivalent to almost $2800, enough to build a really powerful gaming desktop.

jacquesm 7y ago

The 6502 was an enabler if there ever was one. Yes, memory was expensive. But we used only a very small fraction of what we think of as normal today. A typical system would have 2K ram; maybe 4K and if you were rich you'd splurge for the 16K upgrade (but what would you do with so much memory?). In comparision, the 6800 sold for a large multiple of the 6502, and Motorola wouldn't have slashed their prices if not for the 6502. Peddle & Co showed that a CPU did not have to be expensive and that was a game changer. A whole generation of 8 bit machines depended on it: VIC-20, CBM 64, Atari, Acorn Atom, BBC Micro, Apple I, Apple II. None of those would have happened without an affordable CPU to power them. Very few chips can claim such a distinguished legacy, and without showing the market potential for this 'personal computer' thing IBM might not have entered the game at all. Visicalc (which required 32K!) was the breakthrough moment.

agumonkey 7y ago
It's not hard to see the effect on ram price when you have all these tiny personal computers broadening the market. Synergy I say.

whyenot 7y ago

But we also didn't need as much memory. The memory resident portion of CP/M was only 2kb, and that is with some inefficiency because it was necessary to stick with the 8080 instruction set. Just recently there was an article on HN about the passing of Randy Suess, who along with Ward Christensen created the first real BBS. The other part that often isn't recognized, is the work that Irv Hoff did to re-write the memory resident part of CP/M, so that it not only did everything that a operating system needs to do, but could also accept remote connections, deal with a modem, and have some (very rudimentary) security. All in 2kb. That's amazing.

retrac 7y ago
> But we also didn't need as much memory. Another way to look at it is that we only did the things we could do with such little memory. CP/M is a marvel of size-constrained code, but it's also crude in the extreme. A multitasking virtual memory system with a decent file system and an optimizing high-level language compiler -- what most developers would consider the bare minimum for being civilized today -- really relies on having about a megabyte or more of RAM in practice. Consider the web, one of the "killer apps" for computing in general. It's not hard to imagine implementing something like modern packet-switched networking and basic text-only hypertext documents, with a computer which can only do about 100,000 instructions per second. But doing it with only 4 or 16 KB of RAM, though? Just this comment is 843 bytes.
- whyenot 7y ago
  MP/M didn't have virtual memory, but it had everything else on your list, was released in 1979, and had a minimum memory requirement of 32kb. In the mid 1980s, I ran a four line BBS on an Atlas ACS8000-14 running MP/M. I wrote the BBS in DR CBasic. I attempted to port it to Turbo Modela-2, another high level language, but the IDE was so buggy that I eventually gave up (being female and a teenager, there was also a lot of social pressure not do computer stuff).
- wbl 7y ago
  The PDP-11 had 16k.

tyingq 7y ago

The Z80 was around $65 when the $25 6502 came out.

jacquesm 7y ago
And the 6800 was $175, and with the $ exchange rate of the time it was unaffordable here.

pvg 7y ago

such a CPU could have been done in less than 30x30 cm. That's far too big for just the CPU of a home computer. Look at the form factors of the C-64, the Apple ][, the TRS-80, the ZX Spectrum, etc. You had to be able to take it home and plug it into your TV. Something the size of a mini fridge is not really a 'personal' device.

aidenn0 7y ago
30x30 would fit comfortably in those cases, so put it on a riser card and only suffer an extra couple of inches of thickness. Of course the manufacturing price would suffer. I don't see someone making an LSI populated PCB for $25
- pvg 7y ago
  Most of these don't have two sides that are 30cm. The only reason the Apple ][ does is the expansions slots.
  - vidarh 7y ago
    The price of several of them also very much depended on the low cost of assembling small, simple boards. Commodores price points would have been impossible with larger boards with more components, for example. The Apple II was crazy expensive in comparison to many of the other 6502 based computers.

AnimalMuppet 7y ago

Yes, you could build a CPU from a number of MSI chips. But what kind of clock frequency would you get from that kind of a setup? One chip gave you much better performance. A multi-chip CPU would be a toy; a one-chip CPU could do real work (by the standards of the time).

retrac 7y ago
> But what kind of clock frequency would you get from that kind of a setup? One chip gave you much better performance. You might be surprised, because the answer is in the ballpark of 100 MHz when cost is no object. When cost is an object, well even simple designs like the MSI/LSI versions of the PDP-8 were clocked around a 1 microsecond memory cycle, comparable to most early microprocessors. For an extreme example, the Cray-1 was constructed entirely from tens of thousands of discrete NOR gate chips using ECL, and was clocked at 80 MHz. Multi-chip implementations remained standard on big systems into the 1980s. The ability to use bipolar technology over the much slower (at the time) MOS was a major factor, but it also simply wasn't possible to add to a single die sophisticated features like large register sets, pipelining, caches, etc. with the limited transistor budget available in the 1970s. It wasn't until the mid-80s that CMOS microprocessors started to win decisively versus discrete logic for mid-range applications. It lasted even longer on the very high end -- DEC, IBM, Cray, and others, were manufacturing processors constructed out of many MSI/LSI chips well into the 1990s. Even the first POWER processor from IBM consisted of 9 VLSI chips, in 1990. The overwhelming advantage for single-chip processors back then was the low cost. A minicomputer like the PDP-11/20 from all the way back in 1970 constructed out of SSI chips could run circles around the 6502, but the processor for the PDP-11/20 was like $2000 and the 6502 was $25.
- Taniwha 7y ago
  I completely agree with all your points, but would add that we also didn't have the CAD tools to make big stuff, the 6502 was built by cutting litho to make transistors (actually physically cutting different coloured transparent tape and sticking it to a transparent sheet), nowadays I could knock out all fully timed 6502 in a few days with the right tools. Building something the size of a Cray on paper was an amazing achievement