What is a very fast way to scroll through a tilemap in xLIB when your cursor hits the edge of the screen?

I - Cursor size.
N - Where to start drawing tilemap - horiz.
O - Where to start drawign tilemap - vert.
L - cursor's horiz. coord (byte).
M - cursor's vert. coord (byte).

c1-->

CODE

ec1... 8:prgmPTMH Repeat Ans real(8 End Ans->K 8:prgmPTMH prgmPTMK ...c2

-->

CODE

ec1real(12,Ans,LB,MB,B-1+LB,B-1+MBc2

-->

CODE

ec1If K<5:Then L+(K=3)-(K=2)->L M+(K=1)-(K=4)->M If L<0:Then 1->L N-1->N prgmPTMU End If L>10:Then 1->L N+1->N prgmPTMU End If M<0:Then 1->M O-1->O prgmPTMU End If M<8:Then 8->M O+1->O prgmPTMU End Endc2

-->

CODE

ec1real(2,1,N,O,L,M,0,J6,0,J4,Q,0,16/J c2

Well I optimized it a bunch last night.  But I think it could be better...

B - Sprite size (8 or 16)
C - Sprite size - max bytes (2 for 8x8, 1 for 16x16)
H and V - Cursor coordinates.
I and J - The dimensions of the tilemap matrix.
L and M - Where to start drawing the tilemap from the matrix.
Q - The initial picture to get sprites from.

prgmPTMH - Deleted; now using PSE's hilite sub:
c1-->

CODE

ec1real(12,Ans,HB,VB,B+HB-1,B+VB-1c2

-->

CODE

ec1If K<5:Then H+(K=3)-(K=2->H V+(K=1)-(K=4->V (H>C6-1)-(H<->Y If Ans:Then (C6-1)(Ans=1->H L+Y Ans(Ans>=0)-(C6+Ans)I->L prgmAPTMU:End (V>C3-1)-(V<->Y If Ans:Then (C4-1)(Ans=1->V M+Y Ans(Ans>0)-(C4+Ans>J->M prgmAPTMU:End End ...c2

-->

CODE

ec1real(2,1,L,M,I,J,0,C6,0,C4,Q,0,16/Cc2

change

Ans(Ans>=0)-(C6+Ans)I->L

to
Ans(Ans>=0)-I(C6+Ans->L

save one more byte ^^

QuoteBegin-threefingeredguy+Aug 2 2006, 04:07 PM-->

QUOTE (threefingeredguy @ Aug 2 2006, 04:07 PM)

-IC6-IAns+.5(Ans+abs(Ans->L Not smaller, but it may be faster.

I think TFGs would be slower.

yeah,for sure, TFG's is slower,i tested on the calc. /w xlibman's optimisations,it runs at 19 ticks,compared to 22 ticks