Can somebody explain this code

[internetuser2k12]

Hello!

I need a n explanation of

 bYear = (((bYear / 10) << 4) & 0xC0) + ((bYear-(bYear / 10) * 10) & 0b00001111);

see post 12 at http://www.mikroe.com/forum/viewtopic.php?p=84795

He is using byear as a variable which increments from 0 to 99 decimal

He is then converting for eg: decimal 99 to bcd using the above statement

bYear/10 = 99/10 = 0x09 or d9 {integer division)
(bYear / 10) << 4) = 9 << 4 = 0x90 or d144
(bYear / 10) << 4) & 0xC0) = 0x90 & 0xC0 = 0x80 or d128

bYear/10 = 0x09 or d9
(bYear / 10) * 10) = d9 * 10 = d90 or 0x5A
(bYear-(bYear / 10) * 10) = d99 - d90 or 0x63 - 0x5A = d9 or 0x09
((bYear-(bYear / 10) * 10) & 0b00001111) = d9 or 0x09

(((bYear / 10) << 4) & 0xC0) + ((bYear-(bYear / 10) * 10) & 0b00001111) = d128 + d9 or 0x80 + 0x09 = d137 or 0x89

How is he writing 0x89 or d137 instead of 0x99 or dfor year 99. or how is writing d153 for year 99?

i think it should be

 bYear = (((bYear/10)<<4) + (bYear%10));

so that,
(bYear/10) = 99/10 = d9 or 0x09
((bYear/10)<<4) = 0x09 << 4 or d9 << 4 = 0x90 or d144
(bYear%10) = 99%10 = d9 or 0x09
(((bYear/10)<<4) + (bYear%10)) = 0x90 + 0x09 = 0x99 or d144 + d9 = 153

Isn't the second equation right?

Isn't this equation

 (((bYear / 10) << 4) & 0xC0) + ((bYear-(bYear / 10) * 10) & 0b00001111)

wrong for Decimal to BCD Conversion?

 

[Tahmid]

It should be:

bYear = (((bYear / 10) << 4) & 0xF0) + ((bYear-(bYear / 10) * 10) & 0b00001111);

AND-ing with 0xF0 and 0x0F seems unnecessary.


The one you posted is fine. This is fine:

 bYear = (((bYear/10)<<4) + (bYear%10));

Instead of adding, you can also OR:

bYear = (((bYear/10)<<4) | (bYear%10));

 

[internetuser2k12]

OK. Ii got it Tahmid.

 

[FvM]

bYear = (((bYear / 10) << 4) & 0xF0) + ((bYear-(bYear / 10) * 10) & 0b00001111);

Seems like a reminiscence to programming languages that don't know a modulus operator (% in C) like original MS BASIC, or the code has been written by a programmer who din't yet learn about it.

It might be interesting to count the generated number of machine instructions compared to the C standard procedure.

 bYear = (((bYear/10)<<4) + (bYear%10));

It can be still minimized by using assembly code, built-in functions like dec2bcd() are supposed to do.

 

[Tahmid]

So, I tested this with mikroC (selected PIC16F877A as the target device). Here are the results:

bYear = (((bYear / 10) << 4) & 0xF0) + ((bYear-(bYear / 10) * 10) & 0b00001111);

308 cycles.
_____________________________________________________________________________

bYear = (((bYear/10)<<4) | (bYear%10));

bYear = (((bYear/10)<<4) + (bYear%10));

227 cycles.
_____________________________________________________________________________

bYear = DEC2BCD(bYear);

247 cycles.

 

[internetuser2k12]

Good work Tahmid.