|
|
|
| R2012a Documentation → Fixed-Point Toolbox | |
Learn more about Fixed-Point Toolbox |
|
| Contents | Index |
c = bitshift(a, k)
c = bitshift(a, k) returns the value of a shifted by k bits. The input fi object a may be a scalar value or a vector and can be any fixed-point numeric type. The output fi object c has the same numeric type as a. k must be a scalar value and a MATLAB built-in numeric type.
The OverflowMode property of a is obeyed, but the RoundMode is always floor. If obeying the RoundMode property of a is important, try using the pow2 function.
When the overflow mode is saturate the sign bit is always preserved. The sign bit is also preserved when the overflow mode is wrap, and k is negative. When the overflow mode is wrap and k is positive, the sign bit is not preserved.
When k is positive, 0-valued bits are shifted in on the right.
When k is negative, and a is unsigned, or a signed and positive fi object, 0-valued bits are shifted in on the left.
When k is negative and a is a signed and negative fi object, 1-valued bits are shifted in on the left.
This example highlights how changing the OverflowMode property of the fimath object can change the results returned by the bitshift function. Consider the following signed fixed-point fi object with a value of 3, word length 16, and fraction length 0:
a = fi(3,1,16,0);
By default, the OverflowMode fimath property is saturate. When a is shifted such that it overflows, it is saturated to the maximum possible value:
for k=0:16,b=bitshift(a,k);... disp([num2str(k,'%02d'),'. ',bin(b)]);end 00. 0000000000000011 01. 0000000000000110 02. 0000000000001100 03. 0000000000011000 04. 0000000000110000 05. 0000000001100000 06. 0000000011000000 07. 0000000110000000 08. 0000001100000000 09. 0000011000000000 10. 0000110000000000 11. 0001100000000000 12. 0011000000000000 13. 0110000000000000 14. 0111111111111111 15. 0111111111111111 16. 0111111111111111
Now change OverflowMode to wrap. In this case, most significant bits shift off the "top" of a until the value is zero:
a = fi(3,1,16,0,'OverflowMode','wrap'); for k=0:16,b=bitshift(a,k);... disp([num2str(k,'%02d'),'. ',bin(b)]);end 00. 0000000000000011 01. 0000000000000110 02. 0000000000001100 03. 0000000000011000 04. 0000000000110000 05. 0000000001100000 06. 0000000011000000 07. 0000000110000000 08. 0000001100000000 09. 0000011000000000 10. 0000110000000000 11. 0001100000000000 12. 0011000000000000 13. 0110000000000000 14. 1100000000000000 15. 1000000000000000 16. 0000000000000000
bitand | bitcmp | bitget | bitor | bitset | bitsll | bitsra | bitsrl | bitxor | pow2
Learn how to apply early verification to your development process through these technical resources.
How much time do you spend on testing to ensure implementation meets system-level requirements?
| © 1984-2012- The MathWorks, Inc. - Site Help - Patents - Trademarks - Privacy Policy - Preventing Piracy - RSS |
