| Electronics > Projects, Designs, and Technical Stuff |
| Whats the smallest computer you could make? |
| << < (5/6) > >> |
| chris_leyson:
Xilinx's 8-bit PicoBlaze, KCPSM6, fits into 26 logic slices on Spartan-6, Virtex-6 or Virtex-7 and uses 1 block RAM. Older generation KCPSM3 used 96 slices on Spartan3. |
| SiliconWizard:
--- Quote from: joeqsmith on May 31, 2019, 11:47:46 am ---Maybe someone else could chime in on when that term was first introduced. --- End quote --- I think the first chip really akin to a microcontroller was the TMS1000 ( https://en.wikipedia.org/wiki/Texas_Instruments_TMS1000 ) in 1974 (although the TMS1802 in 1971 was already its precursor but much less capable). Then Intel shipped the 8048 in1976. I think both were called something like "single-chip microcomputers" at the time. It looks like the term started to appear in an Intel datasheet for the MCS-48 family in the 80s. Not quite sure about the exact year and if Intel was really the first to coin the term. |
| RJSV:
HI There. I did lots of thinking, and have read some of Alan Turing's 1940's and 1950's work. Attempting to find (full employment) here in 'Silicon Valley' I created and filed for a US Patent on using MECHANICAL LOGIC (2002). The purpose, first was intended for teaching. The simplest summary would be that I worked on creating a mechanical version resembling, more or less, the popular 8-bit 6502 processor. Now, a point I would make, is that you don't strictly need EVERY boolean logic variety to be a computer. Such functions as exclusive OR do not have to be included, as some kind of grade-school requirement... For example, you can build an 'adder', just by committing the results to a block of memory... Say you want 'inverse hyperbolic Secant', well, just pre-calculate the answers, and commit to a memory block. Function is there, now folks can argue the solution 'does not satisfy Turing's Law', and besides, that's ridiculous to use 100 bytes in Read only memory, just to implement an ADDER! Whew, I always have fun with that one... (lol). Thanks |
| David Hess:
Turing machines are a useful standard but not real practical. Useful processors have functions way beyond what a Turing machine requires. For practical implementations, I would say a programmable logic chip plus non-volatile storage so that could get you down to two chips and even be useful. Some programmable logic includes non-volatile storage so that is one chip. This is for a true processor and not just a state machine. At the gate level I am not sure about the smallest size but CPUs can be built with 1000s of gates excluding memory. An online search shows computers built with 100s of transistors. It would be fun to build something using just discrete bipolar injection logic. |
| westfw:
There's a lovely 4chip z80 system that uses a z80, a memory chip, a small bit of glue logic for the z80 to,acccess the memory. And then there’s an avr mixrocontroller that serves as storage,bootloader, and all the peripherals, over on hacaday.io runs cp/m and ismexpandable. The same scheme should work for other cpus as well...... Edit: I left out the link: https://hackaday.io/project/159973-z80-mbc2-4ics-homemade-z80-computer |
| Navigation |
| Message Index |
| Next page |
| Previous page |