Re: ICFP Contest
- From: "Ian Osgood" <iano@xxxxxxxxxxxxx>
- Date: 22 Jul 2006 13:18:41 -0700
Marcel Hendrix wrote:
Dennis Ruffer <druffer@xxxxxxxxxxxxx> writes Re: ICFP Contest
[..]
Here's the output:
LOADING: 9876543210
Wow, I lost faith after an hour and some 10,000,000 um cycles!
I thought somebody said it took 1 Gcycles to get to LOADING, and
I wasn't prepared to wait 10 .. 100 hours (not before being sure
the code is error free!)
Actually, I think I misinterpreted the units reported by cputime;
gforth-fast on my simulator is doing 2 mips.
It only took about 15 minutes to decompress the codex using gforth-fast
on a 1.5GHz PPC.
[..]
Off to find a buddy memory manager...
There is a classic optimization opportunity for LOAD, it takes
no work when it is asked to replace the current array with the
same array. If you don't optimize that, it copies 3 Mbytes for
nothing. I (think I) had only 3000 um-cycles/seconds because of
that.
-marcel
Indeed. If the source is 0, it functions as a JUMP.
Ian
.
- References:
- ICFP Contest
- From: Dennis Ruffer
- Re: ICFP Contest
- From: Dennis Ruffer
- Re: ICFP Contest
- From: Marcel Hendrix
- ICFP Contest
- Prev by Date: Re: SLOC analysis on C-based Forths
- Next by Date: Re: ICFP Contest
- Previous by thread: Re: ICFP Contest
- Next by thread: Re: ICFP Contest
- Index(es):
Relevant Pages
|