[英]Can I set a sequence of bits without unsetting the previous values?
I've got a sequence of bits, say 我说了一些话
0110 [1011] 1111
Let's say I want to set that myddle nybble to 0111
as the new value. 假设我要将myddle nybble设置为
0111
作为新值。
Using a positional masking approach with AND
or OR
, I seem to have no choice but to first unset the original value to 0000
, because if I trying AND
ing or OR
ing against that original value of 1011
, I'm not going to come out with the desired result of 0111
. 使用与
AND
或OR
的位置屏蔽方法,我似乎别无选择,只能先将原始值取消设置为0000
,因为如果我尝试对原始值1011
AND
ing或OR
ing,则我不会预期结果为0111
。
Is there another logical operator I should be using to get the desired effect? 我应该使用另一个逻辑运算符来获得所需的效果吗? Or am I locked into 2 operations every time?
还是我每次都被锁定两次操作?
The result after kindly assistance was: 经过善意的协助,结果是:
inline void foo(Clazz* parent, const Uint8& material, const bool& x, const bool& y, const bool& z)
{
Uint8 offset = x | (y << 1) | (z << 2); //(0-7)
Uint64 positionMask = 255 << offset * 8; //255 = length of each entry (8 bits), 8 = number of bits per material entry
Uint64 value = material << offset * 8;
parent->childType &= ~positionMask; //flip bits to clear given range.
parent->childType |= value;
}
...I'm sure this will see further improvement, but this is the (semi-)readable version. ...我相信这会得到进一步的改进,但这是(半)可读版本。
If you happen to already know the current values of the bits, you can XOR: 如果您碰巧已经知道这些位的当前值,则可以进行XOR:
0110 1011 1111
^ 0000 1100 0000
= 0110 0111 1111
(where the 1100
needs to be computed first as the XOR between the current bits and the desired bits). (其中
1100
需要计算为当前位与所需位之间的XOR)。
This is, of course, still 2 operations. 当然,这仍然是2个操作。 The difference is that you could precompute the first XOR in certain circumstances.
区别在于您可以在某些情况下预先计算第一个XOR。
Other than this special case, there is no other way. 除了这种特殊情况,没有其他方法。 You fundamentally need to represent 3 states: set to 1, set to 0, don't change.
从根本上讲,您需要表示3种状态:设置为1,设置为0,不要更改。 You can't do this with a single binary operand.
您不能使用单个二进制操作数来执行此操作。
You may want to use bit fields (and perhaps unions if you want to be able to access your structure as a set of bit fields and as an int at the same time) , something along the lines of: 您可能需要使用位字段(如果希望能够同时作为一组位字段和int来访问结构,则可能要使用并集),类似于:
struct foo
{
unsigned int o1:4;
unsigned int o2:4;
unsigned int o3:4;
};
foo bar;
bar.o2 = 0b0111;
Not sure if it translates into more efficient machine code than your clear/set... 不知道它是否比您的清除/设置转换成更有效的机器代码...
Well, there's an assembly instruction in MMIX for this: 嗯,MMIX中有一个汇编指令:
SETL $1, 0x06BF ; 0110 1011 1111
SETL $2, 0x0070 ; 0000 0111 0000
SETL rM, 0x00F0 ; set mask register
MUX $1,$2,$1 ; result is 0110 0111 1111
But in C++ here's what you're probably thinking of as 'unsetting the previous value'. 但是在C ++中,您可能会想这就是“重置先前的值”。
int S = 0x6BF; // starting value: 0110 1011 1111
int M = 0x0F0; // value mask: 0000 1111 0000
int V = 0x070; // value: 0000 0111 0000
int N = (S&~M) | V; // new value: 0110 0111 1111
But since the intermediate result 0110 0000 1111 from (S&~M)
is never stored in a variable anywhere I wouldn't really call it 'unsetting' anything. 但是由于
(S&~M)
〜M)的中间结果0110 0000 1111从未存储在任何位置的变量中,所以我不会真正将其称为“未设置”。 It's just a bitwise boolean expression. 这只是一个按位的布尔表达式。 Any boolean expression with the same truth table will work.
具有相同真值表的任何布尔表达式都可以使用。 Here's another one:
这是另一个:
N = ((S^V) & M) ^ A; // corresponds to Oli Charlesworth's answer
The related truth tables: 相关的真值表:
S M V (S& ~M) | V ((S^V) & M) ^ S
0 0 0 0 1 0 0 0 0
* 0 0 1 0 1 1 1 0 0
0 1 0 0 0 0 0 0 0
0 1 1 0 0 1 1 1 1
1 0 0 1 1 1 1 0 1
* 1 0 1 1 1 1 0 0 1
1 1 0 0 0 0 1 1 0
1 1 1 0 0 1 0 0 1
^ ^
|____________________|
The rows marked with '*' don't matter because they won't occur (a bit in V will never be set when the corresponding mask bit is not set). 标有“ *”的行无所谓,因为它们不会发生(当未设置相应的掩码位时,V中的位将永远不会被设置)。 Except for those rows, the truth tables for the expressions are the same.
除了那些行外,表达式的真值表是相同的。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.