Mirror of forum.nim-lang.org

5765 :: How to get left most 8 bit's value from a 32 bit integer ?

• kcvinu (orginal) [2020-01-02T17:32:01+01:00] view original

  Hi all, I have 32 bit integer value and i need to get left most 8 bit from it. Not only left most, actually, i need first 8 bit, then second 8 bit, then third 8 bit. How do i get this ?

---

• miran (orginal) [2020-01-02T18:01:32+01:00] view original

  I would do bitmasking + bitshifting:

  let a = 0b1100_0101
  
  let a_high = (a and 0b1111_00000) shr 4
  echo a_high # => 12 (0b1100)
  Run

---

• kcvinu (orginal) [2020-01-02T20:39:42+01:00] view original

  Thank you b3liever, Though, i did not understand what is happening there, that template worked.

---

• kcvinu (orginal) [2020-01-02T20:40:03+01:00] view original

  Thank you miran.

---

• Stefan_Salewski (orginal) [2020-01-02T20:49:56+01:00] view original

  i did not understand what is happening there,

  There is nothing special. First it has not to be a template, a proc would do. And you have to know only masking and shifting. You shift the bits right, and then mask all but the lower 8 bits. Same as in C. Another way is casting to array of bytes and accessing the bytes as in

  https://forum.nim-lang.org/t/5713#35477

---

• kcvinu (orginal) [2020-01-03T16:11:39+01:00] view original

  Thanks for the reply. Well, array of bytes is a good idea. Is there any speed penalty in that method ?

---

• Stefan_Salewski (orginal) [2020-01-03T16:50:57+01:00] view original

  The optimizing C compiler generates generally very good code for all the cases, but when you really need utmost performance you may inspect the generated assembler code. The cast to array may be a good solution when you need indeed bytes as result, but when you need ints as result, only with the selected bits, then shift and mask may be the better option. Note that you can exchange the order of shift and mask operation, you can first shift and then mask the lower 8 bits, or first mask and then shift. I think first shifting is simpler, because so you can always use the the same bitmask. In C for such operations Union types can be used, but Nim object variants does not allow such re-interpretion of content.

---

• kcvinu (orginal) [2020-01-05T09:47:33+01:00] view original

  Thank you for the detailed reply.

---