double precision sum an array single precision floats assembly language 8087 x87 NPX FPU math coprocessor

sum_array_float.h

double sum_array_float(const float* array, uint16_t count) {
		double sum;
		__asm {
			.8086
			.8087

			mov	cx, count
			jcxz	END				
			lds	si, array		; ds:[si] points to array of floats size cx

			fldz
		ADD_LOOP:
			fadd	dword ptr [si]		; dword is 32 bit single precision float
			add	si, 4			; IEEE float single precision is 4 bytes
			loop	ADD_LOOP
		END:
			fstp	sum
			fwait				; be sure 8087 is done
		}
		return sum;
	}