Re: merits of Lisp vs Python

Jon Harrop wrote:
André Thieme wrote:
Progress is coming exponetially, it is not linear.

Given that "progress" can't be quantified, a statement about its
mathematical evolution isn't very useful...

Exactly

Depending on what function underlies the technology an exponential graph
looks like a straight line on some parts, and therewith linear.

We're going to hit problems in physics pretty soon.

Maybe, maybe not. Chips as they are built today certainly are not
following Moores Law anymore, they have been developing more slowly for
~6 years now. New technology, if anyone comes up with any, need not be
so limited. The physical limits on computation are quite far off.

That is correct. In ca. 5-7 years the best supercomputer on earth will
already have the calculation capabilies like the human brain.

IIRC, Bishop said that the fastest computers could simulate 12 neurons
realistically in real time in 1996. That is over a billion times fewer than
found in the human brain.

There is a big difference here. Andre is just comparing computational
power. You are talking about simulating one type of computer using a
much different one.

There is not evidence that things that work like animal neurons are the
only path to intelligent machines. In fact they're so difficult to
deal with they're almost worth ignoring in my opinion. To achieve
intelligent machines you need another path, that is what has not been
forthcoming, and I don't see anyone coming up with one for a long time.

I don't agree with Andre either. Who knows what the computational
ability of the human brain is? How can it be compared to a computer
when they work in very different ways? Any comparisons are just
guesses. They probably depend entirely on what the "intelligent
program" is doing.

Anyway: we don't need such a strong intelligence to make Lisp compilers
generate faster code. It will happen gradually. Over the coming years
Lisp compilers will generate better and better code. Also Python will
get faster.

That is a very bizarre statement to make. People want to know how fast
language implementations are on a given computer. Saying that Lisp will be
as fast as other languages already are is fruitless.

Perhaps it will be interesting to see what happens when computers are 1,000x
faster and Stalin-compiled Scheme can compile a simple program in under
1sec. I doubt that will cause any radical shift in tool use though.

Maybe it will, maybe it won't. What I'd say though is:-
* The number of programs where performance is a key requirement seems
to be going down.
* There are many further optimizations that can be done of current
tools.

In the OCaml compiler for example there are still ways that the code
generated could be made better. More unboxing could be used for
example, some of the tricks with specialisation from MLton could be
used. In Lisp, many of the same tricks could be used. Also, programs
that make lisp system images into restricted executable programs
(tree-shakers) could be made more sophisticated.

There is not the same amount of potential improvement open to other
languages like C where much of the work has been done.

.

Relevant Pages

  • Re: a dozen cpus on a chip
    ... there are several compilers still available which implement ISO M2. ... By the time the full spec was published most industrial people had moved to other languages for purely commercial reasons. ... Some of the better static analysis tools for Modula2 can see into the program and find very deep faults that would not be found in normal execution. ... It is a common misconception that Lisp is always interpretted. ...
    (sci.electronics.design)
  • Re: How Common Lisp sucks
    ... smattering of scripting languages around the edges). ... using different compilers on every ... SBCL/x86 virtual machine with a program compiled for the clisp virtual ... None of this is meant to detract from lisp. ...
    (comp.lang.lisp)
  • Re: Is octave good to do image processing?
    ... But in any case I consider it untrue that a Lisp program is more ... modern languages and compilers are completely cache unaware. ... compilers that allow you to exploit the full power of your hardware. ... You know what languages hardware drivers are written in by the ...
    (comp.dsp)
  • Re: Why dont we have "C" machines? (was Re: [OT] Re: Should *most* memory be release back to the sys
    ... written either in assembler or other "system programming" languages like ... But now most operating systems are written in C, most compilers ... single task OSes, and multitasking OSes. ... OS (assuming that a lisp machine is the thing where you basically run ...
    (comp.lang.ruby)
  • Re: New Computer Language Shootout?
    ... In Lisp functions, variables and macros usually have longish names. ... compilers, I think. ... learned from the lisp culture and I use it in all the languages I ...
    (comp.lang.lisp)
Loading