Alright, this is weird on my machine...

[STxAxTIC]

I get two different answers at the bottom of this code. Someone else tells me they get the same answer twice. Can I get some community testing? Or someone otherwise set me straight please.

Code: QB64: [Select]

1. DIM z AS _INTEGER64
2. z = 9223372036854775807 ' This is the biggest allowed number of this type.
3. '                         (Only here for reference.)
4.  
5. DIM a AS _INTEGER64
6. DIM b AS _INTEGER64
7. a = 2 ^ 55 - 1 '          This number should be odd.
8. b = 36028797018963967 '   The number above should equal this one.
9. PRINT a
10. PRINT b
11.  

[bplus]

It's the ^ that looses precision.

Code: QB64: [Select]

1. Dim As _Unsigned _Integer64 z, a, b, a1
2. z = 9223372036854775807 ' This is the biggest allowed number of this type.
3. z = 36028797018963967
4. '                         (Only here for reference.)
5. a1 = 1
6. For p = 1 To 55
7.     a1 = a1 * 2
8. Next
9. a = 2 ^ 55 - 1 '          This number should be odd.
10. b = 36028797018963967 '   The number above should equal this one.
11. Print a
12. Print b
13. Print a1 - 1
14.  
15.  

[STxAxTIC]

Sure bplus, it's definitely in the power operator, but I'm just gathering some info as to when we see it versus when we don't.

Here are two screenshots from various people (neither are me) whose systems show the two different cases:

[SMcNeill]

I’m not at the PC currently, but I’d guess it’s a case of intermediary precision.

A and B might be integer64s, but you haven’t specified your constants.

Try:


a = 2&& ^ 55&& -  1&& ‘     This number should be odd.
b = 36028797018963967&& '   The number above should equal this one.

[STxAxTIC]

That fix doesn't do the trick on my computer. I'm curious about both on yours now.

[SMcNeill]

Code: QB64: [Select]

1. DECLARE CUSTOMTYPE LIBRARY ".\fpp"
2.     SUB set_dpfpu 'to toggle to double precision floating point math
3.     SUB set_qbfpu 'to toggle back to what most folks will see with QB64 64-bit default math
4. END DECLARE
5.  
6. set_dpfpu
7.  
8. DIM z AS _INTEGER64
9. z = 9223372036854775807 ' This is the biggest allowed number of this type.
10. '                         (Only here for reference.)
11.  
12. DIM a AS _INTEGER64
13. DIM b AS _INTEGER64
14. a = (2 ^ 55) - 1 '          This number should be odd.
15. b = 36028797018963967 '   The number above should equal this one.
16. PRINT a
17. PRINT b
18.  

[SMcNeill]

Same issue as you see here:

Code: QB64: [Select]

1. DECLARE CUSTOMTYPE LIBRARY ".\fpp"
2.     SUB set_dpfpu 'to toggle to double precision floating point math
3.     SUB set_qbfpu 'to toggle back to what most folks will see with QB64 64-bit default math
4. END DECLARE
5.  
6. DIM x AS _FLOAT, y AS _FLOAT
7.  
8.  
9.  
10. 'Let's print our results without screwing with anything first.
11. x = 5##
12. y = x / 9##
13. PRINT USING "QB64 division       #.####################"; y
14.  
15.  
16. 'Set the double precision math
17. set_dpfpu
18. x = 5##
19. y = x / 9##
20. PRINT USING "QB64 division       #.####################"; y
21.  
22. 'Set the QB64 precision math
23. set_qbfpu
24. x = 5##
25. y = x / 9##
26. PRINT USING "QB64 division       #.####################"; y