Redcoders Frenzy
| The Limited Distance Round |
|---|
Organiser:
Christian Schmidt
Rules:
| Coresize: | 8000 |
|---|
| Max. processes: | 800
|
|---|
| Max. cycles: | 100000
|
|---|
| Read limit: | 800
|
|---|
| Write limit: | 800
|
|---|
| Max. length: | 100
|
|---|
| Min. distance: | 800
|
|---|
| Rounds: | 200
|
|---|
| Points per win: | 3
|
|---|
| Points per tie: | 1
|
|---|
| Points per loss: | 0
|
|---|
You need for this round CoreWin 2.0
Hints:
Using write and read limitations needs totally different strategies. This will be really a round in a not well discovered field.
Well, as Dave figured out in his posting on r.g.c., the read/write limitation in CoreWin is totally different to that what was discussed as a possible pmars extension.
It's like a core inside a core. All values will be wrapped. Let's have a look to the following example. x is the given value of the b-field and y is the cell where the copied jmp #0 will appear.
0000 mov.i 1, x
0001 jmp #0
x y
===========
400 400
401 7601
799 7999
800 0
1200 400
This means, that we can let run for example a coreclear without any problems. And also all other strategies are
possible. Only imps are a bit trickier because the impstep must be smaller than 400.
But we must keep in mind that we are sitting in a 'core inside a core'. For a paper that isn't a problem, they can easily copy itself through the core, but in the case of a scanner or a stone it's more difficult. Because they are usually stationary. But under the r/w limits they wouldn't be able to beat anything. So, what to do? Well, we must replicate them too :) There are two possibilities to do that.
1. Using a silk-style engine using parallel processes
2. Single process replication
For replicating a scanner the second one seems the better choice, while the first one seems better for selfe-splitting stones and clears.
The p-space can be therefor used to syncronize all your copies around the core and/or to detect a damaged copy for self-repairing.
Here is a zip-file with a collection of old 94x (with write limits) entries.
Results:
Roy van Rijn's non-standard paper Sunrise claims first place
followed by Christian Schmidt with a similar non-standard paper
and David Houston with a silk-style paper. Zul Nadzri send the
only p-warrior in this round while Christian Schmidt the only
scanner hopper and Dave Hillis the only two evovled warriors.