1 2 3 4 5 6 7 8 |
function _ftol( f: double) : Integer; cdecl; begin asm lea eax, f fstp qword ptr [eax] end; result := Trunc(f); end; |
Above is a Delphi inline asm version of the ftol (Floating Point to Integer Conversion) function used by our HotPDF Delphi PDF component.
When compiling above Delphi codes in the 64-bit target platform, Delphi will raise below error:
[dcc64 Error] E1025 Unsupported language feature: ‘ASM’
The reason is: The 64-bit Delphi compiler DCC64 does not support mixed Pascal/assembly code.
DCC64.EXE supports 64-bit assembly code, but the routines must be written completely in assembler.
In addition, the 64-bit asm codes have many different than the 32-bit ones, you can not copy the 32-bit asm codes to the 64-bit assembler routines directly, you must re-write them from scratch, I found below codes from the internet:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
function _ftol: Integer; cdecl; // Assumes double value is in FPU stack on entry // Make a truncation to integer and put it into function result var TmpVal: Int64; SaveCW, ScratchCW: word; asm .NOFRAME fnstcw word ptr [SaveCW] fnstcw word ptr [ScratchCW] or word ptr [ScratchCW], 0F00h ;// trunc toward zero, full precision fldcw word ptr [ScratchCW] fistp qword ptr [TmpVal] fldcw word ptr [SaveCW] mov rax, TmpVal end; |
But maybe we do not really need assembly this routine,
1 2 3 4 |
function _ftol(f: double): integer; cdecl; begin Result := Trunc(f); end; |
The trunc function of Delphi just returns the integer part of a floating point number, that’s the conversion from floating point to integer. The speed of this version should be not bad.
Discover more from losLab Software
Subscribe to get the latest posts sent to your email.