/*"Fixed point" means you use ints
that are scaled by a value.A common
example would be using number of pennies
instead of dollars with a float.

Fixed-point used to be much $TX,"faster",HTML="http://en.wikipedia.org/wiki/X87"$,
but modern processors do well with
floats.It also depends on the compiler
and my compiler is poor with floats.

I often use 64-bit ints with upper 32-bits
as int and lower 32-bits as fraction.

See /Demo/SubIntAccess.HC for how
to access upper or lower 32-bits.

For a complete lst of nonstandard compiler
features, see /Doc/HolyC.DD.
*/

U0 Main()
{
F64 t0,f_sum=0,f_val;
I64 i ,i_sum=0,i_val;

i_val= 2.0000002 *0x100000000;
t0=tS;
for (i=1000000000;i;i--)
i_sum+=i_val;
"IntSum\t\t:%.9f Time:%7.3fs\n",i_sum/ToF64(0x100000000),tS-t0;

f_val= 2.0000002;
t0=tS;
for (i=1000000000;i;i--)
f_sum+=f_val;
"FloatSum\t:%.9f Time:%7.3fs\n",f_sum,tS-t0;
'\n';
}

U0 DoIt2()
{
I64 i=0x123456789ABCDEF0;
"i\t\t=%X\n",i;

"i&0xFFFFFFFF\t=%X\n",i&0xFFFFFFFF;
"i>>32\t\t=%X\n",i>>32;

/* Standard int types are declared
with a special compiler feature which
allows a structure to be accessed as a
whole.That's why the i variable can
be accessed normally in addition to
structure member access I64.The actual
intrinsic compiler type is U64i.

public U64i union I64
{
I8i i8[8];
U8i u8[8];
I16 i16[4];
U16 u16[4];
I32 i32[2];
U32 u32[2];
};

It's not quite as great as it seems
because the compiler decides it cannot
place i into a reg, so there is a
penalty.

For a complete lst of nonstandard compiler
features, see /Doc/HolyC.DD.
*/

"i.u32[0]\t=%X\n",i.u32[0];
"i.u32[1]\t=%X\n",i.u32[1];
}

CPURep;
Main;
DoIt2;

//See /Demo/Lectures/FixedPointAdvanced.HC

/*Program Output$HL,0$WW+H,1$FD,1$

$FG$BG$16 Cores 3.500GHz
Int Sum :2000000199.768690240 Time: 0.803s
Float Sum :2000000225.656127040 Time: 3.615s

i =123456789ABCDEF0
i&0xFFFFFFFF =9ABCDEF0
i>>32 =12345678
i.u32[0] =9ABCDEF0
i.u32[1] =12345678
$HL,1$*/