Talk About Network



Register and Login
Nick
Password
Register create new account Sign up is FREE and you can post replies, new topics, bookmark posts and more!
Recover lost password


Gaming > Core War > Working SlowQsc...
Latest [ Topics | Posts ] Archive Post A New Topic Post a Reply
<< Topic < Post Post 1 of 5 Topic 2227 of 2284
Post > Topic >>

Working SlowQscan

by pauldkline@[EMAIL PROTECTED] Mar 4, 2008 at 03:11 PM

A working slowQscan, hopefully with no typo's.

I ran tests with all 16 triplets, as well as 14,12,10,8 & 6 against 6
q4 programs.  The best scores were with 8 and 6.   Each added triplet
gave up 2-3% more losses to each of the opponents.  Apparently if you
are going to be a slow qscanner you better be a short one :-)   At 6
triplets (24 scans and no 2-jmps) it outscored all the q4 programs
except "unheard-of" by 6-10 points.  I ran all possible permutations
using a cycle limit of 5000 against each opponent.

Replacing the qM/qN pairs changed results against individual opponents
but did not make much difference on average, which is to be expected.
Once you go above step 100 there is nothing to be gained (unless
multiple opponents are using exactly the same qscan grid)

I did not try moving pGo to the end, or varying the separation between
the three components, maybe there is something more there.   Certainly
you should not place the paper launch 60-85 locations before the scan
block, since that is where some of the q4's spend half their bombs.

;redcode-94
;name slowQscan
;author P.Kline
;assert CORESIZE == 8000
;1385,927  ; short list of working qM/qN combinations
;1395,932
;1400,940
;1405,931
;1410,956
;2820,1881
;2830,1880
;3290,937
;3295,938
;3300,939
;4230,2343
;4690,455
;4700,454

qM       equ     2820
qN       equ     1881
qStep    equ     111

pStep    equ     (7*3754)
pStep2   equ     (1303+7*1021)
pGo      mov     <pCopy       ,{pCopy
tG       spl     }2           ,}1
tH       spl     0
tI       spl     0
         mov     <pCopy       ,{pCopy
pCopy    spl     1+7+4000     ,1+7
         mov     <1           ,{1
         spl     2000+7       ,7
pPap     spl     @[EMAIL PROTECTED]
           ,>pStep
         mov     }pPap        ,>pPap
         mov     {pPap        ,<1
         spl     @[EMAIL PROTECTED]
           ,>pStep2
         mov.i   #6000-1-2667 ,}2667
         mov.i   >0           ,}0
     for 20
         dat     0,0
     rof
qGo
        sne     qPtr+(qM+0)*(qN+0)+qStep*0 ,qPtr+(qM+0)*(qN
+0)+qStep*1
        seq     qPtr+(qM+0)*(qN+0)+qStep*2 ,qPtr+(qM+0)*(qN
+0)+qStep*3
qJump   jmp     qDec
        sne     qPtr+(qM+1)*(qN+0)+qStep*0 ,qPtr+(qM+1)*(qN
+0)+qStep*1
        seq     qPtr+(qM+1)*(qN+0)+qStep*2 ,qPtr+(qM+1)*(qN
+0)+qStep*3
qAinc   jmp     qDec ,}qDec
        sne     qPtr+(qM-1)*(qN+0)+qStep*0 ,qPtr+(qM-1)*(qN
+0)+qStep*1
        seq     qPtr+(qM-1)*(qN+0)+qStep*2 ,qPtr+(qM-1)*(qN
+0)+qStep*3
qAdec   jmp     qDec ,{qDec
        sne     qPtr+(qM+0)*(qN+1)+qStep*0 ,qPtr+(qM+0)*(qN
+1)+qStep*1
        seq     qPtr+(qM+0)*(qN+1)+qStep*2 ,qPtr+(qM+0)*(qN
+1)+qStep*3
qBinc   jmp     qDec ,>qDec
        sne     qPtr+(qM+0)*(qN-1)+qStep*0 ,qPtr+(qM
+0)*(qN-1)+qStep*1
        seq     qPtr+(qM+0)*(qN-1)+qStep*2 ,qPtr+(qM
+0)*(qN-1)+qStep*3
qBdec   jmp     qDec ,<qDec
        sne     qPtr+(qM-1)*(qN-1)+qStep*0 ,qPtr+
(qM-1)*(qN-1)+qStep*1
        seq     qPtr+(qM-1)*(qN-1)+qStep*2 ,qPtr+
(qM-1)*(qN-1)+qStep*3
qFdec   djn.f   qDec ,qDec

        sne     qPtr+(qM+2)*(qN+0)+qStep*0 ,qPtr+(qM+2)*(qN
+0)+qStep*1
        seq     qPtr+(qM+2)*(qN+0)+qStep*2 ,qPtr+(qM+2)*(qN
+0)+qStep*3
        jmp     qAinc ,}qDec
        sne     qPtr+(qM+1)*(qN+1)+qStep*0 ,qPtr+(qM+1)*(qN
+1)+qStep*1
        seq     qPtr+(qM+1)*(qN+1)+qStep*2 ,qPtr+(qM+1)*(qN
+1)+qStep*3
        jmp     qBinc ,}qDec
        sne     qPtr+(qM+1)*(qN-1)+qStep*0 ,qPtr+(qM
+1)*(qN-1)+qStep*1
        seq     qPtr+(qM+1)*(qN-1)+qStep*2 ,qPtr+(qM
+1)*(qN-1)+qStep*3
        jmp     qBdec ,}qDec

        sne     qPtr+(qM-2)*(qN+0)+qStep*0 ,qPtr+(qM-2)*(qN
+0)+qStep*1
        seq     qPtr+(qM-2)*(qN+0)+qStep*2 ,qPtr+(qM-2)*(qN
+0)+qStep*3
        jmp     qAdec ,{qDec
        sne     qPtr+(qM-1)*(qN+1)+qStep*0 ,qPtr+(qM-1)*(qN
+1)+qStep*1
        seq     qPtr+(qM-1)*(qN+1)+qStep*2 ,qPtr+(qM-1)*(qN
+1)+qStep*3
        jmp     qBinc ,{qDec
        sne     qPtr+(qM-2)*(qN-1)+qStep*0 ,qPtr+
(qM-2)*(qN-1)+qStep*1
        seq     qPtr+(qM-2)*(qN-1)+qStep*2 ,qPtr+
(qM-2)*(qN-1)+qStep*3
        jmp     qFdec ,{qDec

        sne     qPtr+(qM+0)*(qN+2)+qStep*0 ,qPtr+(qM+0)*(qN
+2)+qStep*1
        seq     qPtr+(qM+0)*(qN+2)+qStep*2 ,qPtr+(qM+0)*(qN
+2)+qStep*3
        jmp     qBinc ,>qDec

        sne     qPtr+(qM+0)*(qN-2)+qStep*0 ,qPtr+(qM
+0)*(qN-2)+qStep*1
        seq     qPtr+(qM+0)*(qN-2)+qStep*2 ,qPtr+(qM
+0)*(qN-2)+qStep*3
        jmp     qBdec ,<qDec
        sne     qPtr+(qM-1)*(qN-2)+qStep*0 ,qPtr+
(qM-1)*(qN-2)+qStep*1
        seq     qPtr+(qM-1)*(qN-2)+qStep*2 ,qPtr+
(qM-1)*(qN-2)+qStep*3
        jmp     qFdec ,<qDec

        sne     qPtr+(qM-2)*(qN-2)+qStep*0 ,qPtr+
(qM-2)*(qN-2)+qStep*1
        seq     qPtr+(qM-2)*(qN-2)+qStep*2 ,qPtr+
(qM-2)*(qN-2)+qStep*3
        djn.f   qFdec ,qDec
        jmp     pGo
     for 4
         dat     0,0
     rof
null     dat     0,0
qSteps  dat     qStep*2-1 ,qStep*2
qDec    mul.x   #qM     ,#qN

        add.f   qDec    ,qPtr
        sne     *qPtr   ,@[EMAIL PROTECTED]
        add.f   qSteps  ,qPtr

        sne     null    ,@[EMAIL PROTECTED]
        mov.x   qPtr    ,qPtr
qPtr    mov     0       ,{qStep
        mov     null    ,>qPtr
        jmn.f   qPtr    ,@[EMAIL PROTECTED]
        jmp     pGo

        end     qGo




 5 Posts in Topic:
Working SlowQscan
pauldkline@[EMAIL PROTECT  2008-03-04 15:11:09 
Re: Working SlowQscan
Roy <roy.van.rijn@[EMA  2008-03-05 01:04:03 
Re: Working SlowQscan
pauldkline@[EMAIL PROTECT  2008-03-05 05:55:01 
Re: Working SlowQscan
Roy <roy.van.rijn@[EMA  2008-03-05 08:17:09 
Re: Working SlowQscan
pauldkline@[EMAIL PROTECT  2008-03-05 11:42:25 

Post A Reply:
  Go here to Signup

AddThis Feed Button


About - Advertising - Contact - Frequently Asked Questions - Privacy Policy - Terms of Use - Signup

Contact
tan12V112 Sat May 17 1:37:54 CDT 2008.