[racket-dev] Inline caching (was Re: my '312' this semester, how we compare to others)

From: Jos Koot (jos.koot at telefonica.net)
Date: Wed May 4 17:42:26 EDT 2011


> -----Original Message-----
> From: dev-bounces at racket-lang.org 
> [mailto:dev-bounces at racket-lang.org] On Behalf Of Matthias Felleisen
> Sent: 04 May 2011 22:58
> To: Tony Garnock-Jones; D Herring
> Cc: dev List
> Subject: Re: [racket-dev] Inline caching (was Re: my '312' 
> this semester,how we compare to others)
> As I was reading a paper, I encountered the word 'cache' and 
> remembered that I had wanted to implement a caching version 
> of the central method in my central data structure. That took 
> me about 10 minutes, and I now have good times for the game: 
> with contracts: 1.8s per game 
> without ctrcts: 1.2s per game 
> I am now within 20% of the fastest Java implementation in my 
> class. That's good. 
> -- Matthias
> p.s. Okay, I'll say it: algorithms matter, too. 

That to me seems most important.
Of course inplementation of an algortithm requires wisdom too:

Is cashing of data faster than recomputing them?
How do we deal with shortage of RAM?
How do we optimize software? (nowadays surely not by hand)
Do software devellopers inspire hardware designers, or may/should it be the
orher way around?
There is a tendency to separate programmers from the hardware equipment they
are using. Is this wise?

I think hardware designers must understand programmers and vice versa.
Most important, both parties must understand their customers.

mho, Jos

> _________________________________________________
>   For list-related administrative tasks:
>   http://lists.racket-lang.org/listinfo/dev

Posted on the dev mailing list.