Python to surpass C performance by 2030

Peter Hansen peter at engcorp.com
Tue Sep 3 14:17:01 CEST 2002


Dan Bishop wrote:
> Erno Kuusela <erno-news at erno.iki.fi> wrote in message news:<kuptw4o65q.fsf at lasipalatsi.fi>...
> 
>>just for fun i ran pystone under python versions since 1.5.2 to
>>current cvs. here are the results:
>>
>>[benchmark results snipped]
>>
>>since 1.5.2 is about 4 years old, the improvement per year is about
>>13.2% on average. if we assume that 1.5.2 is 50 times slower than
>>C (probably conservative?), then at 13.2% per year since 1998 it
>>will add up to 53x improvement in 32 years.
>>
>>i wonder if this would be a good application for the time machine...
>>
>>learn python today, and in 28 years you will have a job writing
>>the speed-critical parts of c programs in python :)
> 
> 
> Before that could ever happen, the Python interpreter would have to be
> written in something other than C.

Clearly if it will be 53x faster in 32 years even if written in C,
then it should be rewritten in Python for another 53x speedup!

Imagine the possibilities.  You could even implement another virtual
machine underneath the current one, to increase speed even more,
provided *that* one was executing Python bytecode as well.  Scary
really.  I understand this plan was discussed at a recent PSU mee




More information about the Python-list mailing list