[CLOSED] Optimization at function TFPReaderBMP.ShiftCount

[lagprogramming]

packages/fcl-image/src/fpreadbmp.pp has

Code: Pascal  [Select][+][-]

1. function TFPReaderBMP.ShiftCount(Mask : longword) : shortint;
2. var tmp : shortint;
3. begin
4.   tmp:=0;
5.   if Mask=0 then
6.   begin
7.     Result:=0;
8.     exit;
9.   end;
10.  
11.   while (Mask mod 2)=0 do { rightmost bit is 0 }
12.   begin
13.     inc(tmp);
14.     Mask:= Mask shr 1;
15.   end;
16.   tmp:=tmp-(8-popcnt(byte(Mask and $FF)));
17.   Result:=tmp;
18. end;

The following patch replaces "while (Mask mod 2)=0 do" with the faster "while (Mask and 1)=0 do" line.

[Zvoni]

Addition:
1) Line 16+17 can be coalesced into one line
2) Why use a tmp-var in the first place?

[RayoGlauco]

My own version may be like this (I agree Zvoni's opinion):

Code: Pascal  [Select][+][-]

1.     function TFPReaderBMP.ShiftCount(Mask : longword) : shortint;
2.     begin
3.       Result:=0;
4.       if Mask=0 then
5.         exit;
6.      
7.       while (Mask and 1)=0 do { rightmost bit is 0 }
8.       begin
9.         inc(Result);
10.         Mask:= Mask shr 1;
11.       end;
12.       Result:= Result-(8-popcnt(byte(Mask and $FF)));
13.     end;

[AlexTP]

Posted and applied:
https://gitlab.com/freepascal.org/fpc/source/-/issues/40590