The Computer and the Mind
Comparisons and thinking about the relationship between computers and the human brain as it pertains to thinking and creativity.
The Computer and the Mind
The problem with the 16-bit micros in the early 80s — such as the Motorola 68000 (used on the Mac) and the Intel 8086 (used for IBM PC compatibles) was not just cycle time but also the fundamental difference between being able to stick “hardware like functions” (such as a byte code interpreter, graphics painting, music, etc.) into a microcode... See more
Just a moment...
A really interesting design was the first ARM — which was a pretty clean RISC and tidy in size. It could have been used as an emulator by wrapping it with fast instruction memory, but wasn’t. I think this was a “point of view” disconnect. It was a very good design for the purpose of its designers, and there wasn’t enough of a VHLL culture to see... See more
Just a moment...
The idea of microcode was invented by Maurice Wilkes, a great pioneer who arguably made the earliest programmable computer — the EDSAC (pace Manchester Baby). The idea depends partly on the existence of a “large enough” memory that is much faster (3–10 times) than the 1st level RAM of the computer.
A milestone happened when the fast memory for... See more
A milestone happened when the fast memory for... See more
Just a moment...
"the question of whether a computer can think is no more interesting than the question of whether a submarine can swim."
- edsger dijkstra
(I can tell that you know this:) The problem with trying to emulate without a separate instruction memory is that the fetches are both (a) slow from the main memory, and (b) use memory cycles that the emulated mechanism needs to use. What is wanted is to not bother main memory for the emulation routine, and to be able to execute 4, 5 or more... See more
Just a moment...
One obvious source of information that a lot of people are working towards harnessing is video. According to a random site on the Web, about 500 hours of video footage are uploaded to YouTube *per minute*. This is a ridiculous amount of data, much more than is available as text on the entire internet. It’s potentially a much richer source of... See more
There Are No New Ideas in AI… Only New Datasets
Video as the panacea to good training data is an interesting thought, but that depends on how you evaluate the quality of the videos’ content. Any knowledge about physical spaces or culture will be limited to that subset of people who post online and the spaces they inhabit. It may lead the AI down a path of hallucination, since image-based media on the internet is largely performative.
Up to a point, an instruction cache can act like a microcode memory for emulating VHLLs. The keys are for it (a) to be large enough to hold the inner loops of the interpreter, (b) to not be flushed spuriously, and (c) for the machine instructions to execute quickly compared to the cache memory cycle.
Just a moment...
The biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin.
hyperthesis • The Bitter Lesson
The methods need to require more computational resources rather than particularly clever encodings of human specialist knowledge. Resources can be time (to run computer microcode instructions) or space (memory to hold data).
At the beginning I said I didn't think the Alto Architecture is better than current approaches. The main reasons are the things that Alto did not have:
As a consequence of the... See more
- No virtual memory or any memory protection
- No executive mode, to support a trusted operating system or contain untrusted code
- No virtualization capability
- No parallelism
- No caches
As a consequence of the... See more