US20070120718A1

US20070120718A1 - Method and Apparatus for Binary Number Conversion

Info

Publication number: US20070120718A1
Application number: US11/671,063
Authority: US
Inventors: Peng-Hua Wang
Original assignee: Individual
Current assignee: XIAHOU HOLDINGS LLC
Priority date: 2003-07-10
Filing date: 2007-02-05
Publication date: 2007-05-31
Also published as: US7191200B2; US20050010622A1

Abstract

The method and apparatus use two inequalities to determine whether an estimated value obtained from conventional method and a correct value obtained from ideal conversion is identical. When those values are the same, the estimated value is not corrected; otherwise, according to the difference between those values, one is added or subtracted from the estimated value to obtain the correct result.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application is a continuation of U.S. application serial number Ser. No. 10/616,646, filed Jul. 10, 2003, the contents of which are incorporated by reference herein.

BACKGROUND OF THE INVENTION

1. Field of Invention
The present invention relates to digital data processing. More particularly, the present invention relates to a method and an apparatus for binary number conversion.
2. Description of Related Art
Conversions between fixed-point binary numbers occur frequently in digital systems. When processing bits of two electronic devices in the same system are different, for example, a signal input into a video card is 5 bits but internal operation of the video card is 8 bits, the data processing between the input and internal operation needs bits conversion. These data in the video card usually represents colors, brightness, and coordinates. If some mistakes occur during bits conversion, it severely obstructs the normal operation of the video card.
The foregoing bits conversion of a binary number converts an m-bit binary number to an n-bit binary number. An unsigned m-bit binary number is able to represent an integer from 0 to 2^m−1. An ideal conversion from an m-bit binary number to its n-bit counter part is to map 0 (the smallest value represented by m bits) to 0 (the smallest value represented by n bits), and to map 2^m−1 (the largest value represented by m bits) to 2ⁿ−1 (the largest value represented by n bits). Besides, other values of the m-bit binary number also respectively and individually correspond to other values of the n-bit binary number, and it is a one-to-one ideal corresponding relation.
This one-to-one ideal corresponding relation is expressed by the equation $\begin{matrix} y = f (x) = [x \cdot \frac{2^{n} - 1}{2^{m} - 1} + 0.5] & (1) \end{matrix}$
where x is a value of an m-bit binary number, and y is a correct value of an n-bit binary number converted from m-bit binary number. The addition of 0.5 thereto means “round to nearest”. The symbol [I] represents the floor equation or Gauss equation whose value is the largest integer less than I. For example, a value of [3.4] is 3, and a value of [4.6] is 4. Combination of [ ] and +0.5 therein can therefore accomplish rounding operation; i.e. a value of [3.4+0.5] is still 3, but a value of [4.6+0.5] becomes 5. From the foregoing description, the equation (1) linearly maps x=0 (the smallest value represented by m bits) to y=0 (the smallest value represented by n bits), and x=2^m−1 (the largest value represented by m bits) to y=2ⁿ−1 (the largest value represented by n bits). Moreover, other values of x also respectively and individually map to other corresponding correct values of y.
However, modern digital systems do not use this ideal conversion method to convert a binary number between different bits, but rather use a method as illustrated in the following with reference to FIG. 1A to quickly convert a binary number between different bits without correctly rounding.
FIG. 1A is a schematic view of a conventional conversion method of a binary number between different bits. As illustrated in FIG. 1A, when an m-bit binary number 102 is converted to an n-bit binary number 104, the conversion method first supposes
n=mq+r,
where 0≦r<m, and m, n, q, r are positive integers. (2)
Then the binary number 102 is duplicated q times to obtain an m*q bits binary number. The m*q bits binary number is regarded as a most significant bits number 108 of the binary number 104. Further, the most significant r bits of the binary number 102 are copied to be a least significant bits number 106 of the binary number 104. The conventional method thus duplicates and combines the binary number as foregoing to obtain quickly the converted n-bit binary number 104 without a real numerical operation.
In FIG. 1A, a value of the binary number 102 is x, and a value of the binary number 104 converted thereby is y′, the y′ is an estimated value converted from the binary number 102, and is different from the correct value y. The conventional method obtains the binary number 104 without the real numerical operation, and the estimated value y′ can be represented by equation $\begin{matrix} y^{'} = x \cdot (\sum_{i = 1}^{q} 2^{r + m (q - 1)}) + [\frac{x}{2^{m - r}}] & (3) \end{matrix}$
However, the value y′ obtained from equation (3) of the conventional conversion method is an estimated value, and may not be equal to the value y obtained from equation (1) of ideal conversion method. The values y and y′ of some values x respectively obtained from these two method are different. The following description cites an instance to interpret this inaccuracy, and uses a table to explain it in detail.
A 5-bit binary number 112 in FIG. 1B is converted to an 8-bit binary number 114 by using the conventional conversion method as illustrated in FIG. 1A. The 5-bit binary number 112 is composed of α, β, γ, δ, and ε, the 5 bits are arranged in order from the most significant bit to the least significant bit.
In FIG. 1B, the m and n are 5 and 8, respectively. The 5-bit binary number 112 needs to be duplicated once to be a most significant bits number 118 of the 8-bit binary number 114 during conversion. And the most significant r=3 bits α, β, and γ of the 5-bit binary number 112 are duplicated as a least significant number 116 of the 8-bit binary number 114. The foregoing conventional conversion method thus can obtain the 8-bit binary number 114, and α, β, γ, δ, ε, α, β, and γ thereof are arranged from the most significant bit to the least significant bit, as illustrated in FIG. 1B.
The description thereinafter describes several actual values x of the 5-bit binary number 112 are converted as several estimated values y′ of the 8-bit binary number 114, and compared with the correct values y thereof obtained from the ideal conversion method in Table. 1 to represent concretely errors made by the conventional conversion method. The 5-bit binary number 112 can used to represent 32 numbers from 0 (00000) to 31 (11111). These 32 numbers are respectively converted by ideal and conventional conversion method to obtain 32 8-bits numbers whose minimum value is (00000000) and maximum value is 255 (11111111).

In Table. 1, a value x is a value of the original 5-bit binary number. A corrected value y is a value obtained by equation (1) of the ideal conversion method. An estimated value y′ is a value obtained by equation (3) of the conventional conversion method. A difference value e is a value of (y-y′), and the difference value e is represented by

\begin{matrix} e = y - y^{'} = [x \frac{2^{3} - 1}{2^{5} - 1} + 0.5] - [\frac{x}{2^{2}}] & (4) \end{matrix}

TABLE 1


A comparison between the values of 5-bit binary number and
8-binary number converted thereby.

	x	y	y′	e

0	0	0	0
1	8	8	0
2	16	16	0
3	25	24	1
4	33	33	0
5	41	41	0
6	49	49	0
7	58	57	1
8	66	66	0
9	74	74	0
10	82	82	0
11	90	90	0
12	99	99	0
13	107	107	0
14	115	115	0
15	123	123	0
16	132	132	0
17	140	140	0
18	148	148	0
19	156	156	0
20	165	165	0
21	173	173	0
22	181	181	0
23	189	189	0
24	197	198	−1
25	206	206	0
26	214	214	0
27	222	222	0
28	230	231	−1
29	239	239	0
30	247	247	0
31	255	255	0

If the difference value e is equal to zero, the estimated value y′ and the correct value y are identical. The estimated value is the correct result, and additional modification is unnecessary. But if the difference value e is not equal to zero, the estimated value y′ is different from the correct value y, and some mistakes occur during bits conversion of a binary number between different bits.
As illustrated in Table. 1, when the value of the original 5-bit binary number 112 is 3, 7, 24, or 28, the estimated value y′ of the 8-bit binary 114 obtained by conventional conversion method is different from the correct value y obtained by ideal conversion method, respectively. When x is 3 or 7, the estimated value y′ is more than the correct value y with one; when x is 24 or 28, the estimated value y′ is less than the correct value y with one. The conventional conversion method therefore generates mistakes during conversion of a binary number between different bits when x is 3, 7, 24, or 28.
For digital data processing, a mistake during conversion of a binary number between different bits causes very serious problems. For example, these incorrect estimated values obtained by conventional conversion method generally are input values for a subsequent operation, and after many operations, the inaccuracy between the results of operating with incorrect estimated values and correct values becomes very large.
In an example of a video card, digital data usually represents colors, brightness, and/or coordinates. If mistakes occur during bits conversion, deviations of colors and coordinates may be caused. The deviation of colors makes colors of an image outputted from the video card incorrect, and the deviation of coordinates makes a graph plotted referring to these coordinates incorrect. Furthermore, the video card uses the converted values to query a table, or uses the converted values to be a reference coordinates to query colors. These queries generate mistakes due to these incorrect estimated values, and thus severely obstruct the normal operation of the video card.

SUMMARY OF THE INVENTION

It is therefore an objective of the present invention to provide a method and an apparatus for binary number conversion that satisfies the need to solve the problem of conversion of a binary number between different bits.
In accordance with the foregoing and other objectives of the present invention, a method and an apparatus are described. The method and apparatus of the invention use two inequalities to determine whether an estimated value obtained from conventional method and a correct value obtained from ideal conversion is identical. When those values are the same, the estimated value is not corrected; otherwise, according to the difference between those values, one is added or subtracted from the estimated value to obtain the correct result.
In one preferred embodiment of the present invention, an m-bit first binary number in a digital data processing system is converted to an n-bit second binary number, wherein n=m*q+r, 0≦r<m, and m, n, q, r are positive integers.
The first binary number is duplicated from the most significant bit to the least significant bit thereof to form a n-bit third binary number, in which r bits of the first binary number duplicated at the last time are defined as a first number, and remaining bits of the first binary number without the first number are defined as a second number.
Then the first number and the second number are swapped to form a fourth binary number. A value of the second binary number is subtracted from a value of the fourth binary number to obtain a difference value. When the difference value is larger than or equal to 0.5*(2^m−1), one is added to the first number of the third binary number to obtain the second binary number; when the difference value is less than −0.5*(2^m−1), one is subtracted from the first number of the third binary number to obtain the second binary number; when the difference value is not larger than or not equal to 0.5*(2^m−1), and not less than −0.5*(2^m−1), the third binary number is equal to the second binary number.
In another preferred embodiment of the present invention, an original binary number, an original bit quantity m, and a converted bit quantity n are input into a determination unit, and the determination unit calculates q and r by n=m*q+r, and 0≦r<m.
The determination unit sends the original binary number to a bits replicator. The bits replicator duplicates the original binary number q times, and obtains a binary number composed of m*q bits. Moreover, the determination unit further sends the original binary number to a mask. The mask masks the unnecessary bits of the original binary number according to r and then obtains a binary number composed of r bits.
A bits swapper swaps the r-bit most significant bits number and the (m−r) bits least significant bits number to form a swapped binary number. Positive and negative signs are respectively attached to the swapped binary number and the original binary number, and then they added together by an adder. A result obtained from adder is sent to the 2-level comparator. An incrementor/decrementor modifies the r-bit binary number received from the mask according to the comparison result.
Subsequently, an r-bit binary number modified in the incrementor/decrementor is sent to a combination unit to be the least significant bits of the result binary number. At the same time, the m*q-bit binary number obtained from the bits replicator is also sent to the combination unit to be the most significant bits of the result binary number. The combination unit therefore combines the two binary numbers provided by the bits replicator and the incrementor/decrementor, and outputs a correct converted binary number from an output port.
In another preferred embodiment of the present invention, a method and an apparatus of the invention is utilized in a video card. The foregoing second binary number is represented colors or coordinates.
In conclusion, the invention modifies an incorrect result obtained from the conventional conversion method by a simple method to obtain the correct result, without complicated rounding. The invention also uses a difference value between the original binary number and the swapped binary number thereby to compare with a predetermined value. According to the difference value, the simple 2-level comparison method of the invention therefore modifies the estimated value from the convention conversion, finally obtaining the correct result.
In another aspect, the invention can be implemented with only the bits duplication and combination of binary numbers, and a simple numerical comparison step. In digital data processing systems, such as video cards, the processing speed is faster following the great progress of the science and technology. Methods and apparatuses able to quickly and correctly accomplish conversion between different bits are therefore needed. The method of the invention no only avoids the deviations of colors and coordinates in the conventional video card, but also does not slow the processing speed of the video card, so as to make the video card operate normally.
It is to be understood that both the foregoing general description and the following detailed description are examples, and are intended to provide further explanation of the invention as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS

These and other features, aspects, and advantages of the present invention will become better understood with regard to the following description, appended claims, and accompanying drawings where:
FIG. 1A is a schematic view of a conventional conversion method of a binary number between different bits;
FIG. 1B is a schematic view of a 5-bit binary number converted to an 8-bit binary number by using the conventional conversion method in FIG. 1A;
FIG. 2 is a schematic view of x and x′ in accordance with one embodiment of the invention;
FIG. 3 is a flow chart of the method of the invention; and
FIG. 4 is a schematic view in accordance with one embodiment of the invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Reference will now be made in detail to the present preferred embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers are used in the drawings and the description to refer to the same or like parts.
The present invention provides a method and an apparatus for binary number conversion to solve the problem of conversion of a binary number between different bits.
The method and apparatus of the invention use two inequalities to determine whether an estimated value obtained from conventional method and a correct value obtained from ideal conversion is identical. When those values are the same, the estimated value is not corrected; otherwise, according to the difference between those values, one is added or subtracted from the estimated value to obtain the correct result.
The following descriptions explain the invention by the m-bit binary number 102 and the n-bit binary number 104 converted thereby in FIG. 1A. For simple and clear explanation, the q of the equation (2) is set to 1 to make a bit quantity relation between the two binary numbers become n=m+r. However, if the bit quantity relation between the two binary numbers is q being an integer more than one, the invention also can convert a binary number to obtain the correct result.
A binary number 202 is composed of m bits, and a value thereof is x. Therefore, a correct value of a binary number thereof converted by the equation (1) of the ideal conversion method is $\begin{matrix} y = f (x) = [x \cdot \frac{2^{n} - 1}{2^{m} - 1} + 0.5] & (1^{'}) \end{matrix}$
The binary number 202 also can be converted to a binary number by the equation (3) of the conventional conversion method, and the estimated value thereof is $\begin{matrix} y^{'} = 2^{r} x + [\frac{x}{2^{m - r}}] & (5) \end{matrix}$
A difference value between the correct value y and the estimated value y′ obtained from the two conversion method is $\begin{matrix} e = y - y^{'} = [x \frac{2^{r} - 1}{2^{m} - 1} + 0.5] - [\frac{x}{2^{m - r}}] & (6) \end{matrix}$
Divide the binary number 202 into two portions as a most significant bits number 212 and a least significant bits number 222 illustrated as FIG. 2. The most significant bits number 212 is composed of r bits, and a value thereof is a; the least significant bits number 222 is composed of (m−r) bits, and a value thereof is b. The value x of the binary number 202 can be represented by a and b as
x=2^m−r a+b,where 0≦a<2′ and 0≦b<2^m−r (7)
Equation (7) is substituted into x of equation (6) to simplify equation (6) as $\begin{matrix} e = [\frac{a (- 2^{m - r} + 1) + b (2^{r} - 1)}{2^{m} - 1} + 0.5] & (8) \end{matrix}$
If the difference value e between the correct value y and the estimated value y′ is not equal to 0, the difference value e must be more than 0 or less than 0, and the absolute value of the difference value e is the integer 1. For easily simplification, the invention selects two conditions: e≧1 and e<0 to represent some conditions of e being not equal to 0, so as to make right sides of two inequalities in these two conditions only different from positive and negative signs thereof. Thus the following two inequalities are obtained
a·(−2^m−r+1)+b·(2^r−1)≧0.5·(2^m−1) (9)
a·(−2^m−r+1)+b·(2^r−1)<0.5·(2^m−1) (10)
The most significant bits number 212 and the least significant bits number 222 of the binary number 202 are swapped to form a new binary number 204. A value of the binary number 204 is x′, the most significant bits number 214 thereof is composed of (m−r) bits, and a value thereof is b; the least significant bits number 224 thereof is composed of r bits, and a value thereof is a. The value x′ of the binary number 204 can be represented as
x′=2^r b+a (11)
Then equation (7) and equation (11) are substituted into x and x′ of inequalities (6) and (7) to obtain two simplified inequalities
x′−x≧0.5·(2^m−1) (12)
x′−x<0.5·(2^m−1) (13)
The invention uses the two inequalities to determine whether an original binary number generates an incorrect result after the conventional conversion. First, a value x of the original binary number and a value x′ of the most and least significant bits number swapped thereof are substituted into the inequalities (12) and (13). When the inequality (12) is true, e=1, and the estimated value y′ is one less than the correct value y, so one is added to the estimated value y′ to obtain the correct result. In another aspect, when the inequality (13) is true, e=1, and the estimated value y′ is one more than the correct value y, so one is subtracted from the estimated value y′ to obtain the correct result. If the equalities (12) and (13) both are not true, e=0, the estimated value is equal to the correct value y, and no further modification is needed. The invention utilizes the conventional conversion to convert quickly a binary number between different bits and cooperates inequalities to modify the converted binary number, thus quickly and correctly obtains the converted result.
FIG. 3 is a flow chart of one embodiment of the invention. The symbols m, n, x, q, r, a, b, and y′ are the same as in the foregoing description. The m, n, and x are input at the beginning in a step 301. The equation (2), equation (3) and equation (7) are then used to calculate the q, r, a, b, and y, respectively, in step 302. The a, b, and r obtained in step 302 are brought into the equation (11) to calculate the x′ in a step 303.
Afterward, x and x′ are compared by the 2-level steps 304-307. First, x and x′ are substituted into the inequality (12) in a step 304. If a result thereof is true, then a step 305 is carried out to add one to the estimated value y′ to obtain the correct value y. Otherwise, if the result of the inequality (12) is wrong, then a step 306 is carried out, and x and x′ are substituted into the inequality (13). If a result of the inequality (13) is true, then a step 307 is carried out to subtract one from the estimated value y′ to obtain the correct value y. If the result of the inequality (13) is still wrong, then a step 308 is carried out; at this time the estimated value y′ is equal to the correct value y, and no further modification is needed.
FIG. 4 is a schematic view of another preferred embodiment of the invention, and the following descriptions also refer to the flow chart in FIG. 3 to explain the preferred embodiment. Step 301 in FIG. 3 is first carried out to input an original binary number, an original bit quantity m and a converted bit quantity n into a determination unit 401 from an input port. Then the determination unit 401 is in charge of processing the step 302 to calculate q and r.
The determination unit 401 sends the original binary number to a bits replicator 402. The bits replicator 402 duplicates the original binary number by q times, and obtains a binary number composed of m*q bits. Moreover, the determination unit 401 further sends the original binary number to a mask 403. The mask 403 masks the unnecessary bits of the original binary number according to the r and then obtains a binary number composed of r bits.
A bits swapper 404 swaps the r-bit most significant bits number and the (m−r) bits least significant bits number to form a swapped binary number. Positive and negative signs are respectively attached to the swapped binary number and the original binary number, and then they are added together by an adder 405. A result obtained from adder 405 is sent into 2-level comparator 406, and the steps 304 and 306 in FIG. 3 are carried out. An incrementor/decrementor 407 modify the r-bit binary number received from the mask 403 according to the comparison result of the steps 304 and 306 in FIG. 3.
Subsequently, an r-bit binary number modified in the incrementor/decrementor 407 is sent to a combination unit 408 to be the least significant bits of the result binary number. At the same time, the m*q-bit binary number obtained from the bits replicator 402 is also sent to the combination unit 408 to be the most significant bits of the result binary number. The combination unit 408 therefore combines the two binary numbers provided by the bits replicator 402 and the incrementor/decrementor 407, and outputs a correct converted binary number from an output port.
In conclusion, the invention modifies an incorrect result obtained from the conventional conversion method by a simple method to obtain the correct result, without complicated rounding. The invention also uses a difference value between the original binary number and the swapped binary number thereby to compare with a predetermined value. According to the difference value, the simple 2-level comparison method of the invention therefore modifies the estimated value from the convention conversion, and finally obtains the correct result.
In another aspect, the invention can be implemented with only the bits duplication and combination of binary numbers, and a simple numerical comparison step. Digital data processing systems, such as video cards, the processing speed is faster following the great progress of the science and technology. Methods and apparatuses able to accomplish quickly and correctly conversion between different bits are therefore needed. Binary numbers in the invention usually represents colors or coordinates in the video card. The method of the invention no only avoids the deviations of colors and coordinates in the conventional video card, but also does not slow the processing speed of the video card, so as to make the video card operate normally.
It will be apparent to those skilled in the art that various modifications and variations can be made to the structure of the present invention without departing from the scope or spirit of the invention. In view of the foregoing, it is intended that the present invention cover modifications and variations of this invention provided they fall within the scope of the following claims and their equivalents.

Claims

1. A method for converting an input digital data having m bits into an output digital data having n bits, wherein n=m*q+r, 0≦r<m, and m, n, q, r are positive integers, the method comprising:

duplicating the input digital data from most significant bit to least significant bit thereof to form a first intermediate digital data having n bits, wherein r bits of the input digital data duplicated at a last time are defined as a first number, and remaining bits of the input digital data without the first number are defined as a second number;

swapping the first number and the second number to form a second intermediate data;

subtracting a value of the input digital data from a value of the second intermediate data to obtain a difference value; wherein

when the difference value is larger than or equal to 0.5*(2^m−1), adding one to the first number of the intermediate digital data to obtain the output digital data;+

when the difference value is less than −0.5*(2^m−1), subtracting one from the first number of the intermediate digital data to obtain the output digital data; and

when the difference value is not larger than or not equal to 0.5*(2^m−1), and not less than −0.5*(2^m−1), the intermediate digital data is equal to the output digital data.

2. The method of claim 1, wherein the output digital data represents a color performed by a video card.

3. The method of claim 1, wherein the output digital data represents a coordinate performed by a video card.

4. The method of claim 1, wherein the first number has r bits, and the second number has (m−r) bits.

5. A method for converting an m-bit input digital data having m bits into an output digital data having n bits, wherein n=m*q+r, 0≦r<m, and m, n, q, r are positive integers, the method comprising:

swapping the first number and the second number to form a second intermediate digital data;

obtaining a difference value between a value of the second intermediate digital data and a value of the first digital data; and

modifying the first number of the first intermediate digital data by a rule referring the difference value to obtain the output digital data.

6. The method of claim 5, wherein the output digital data represents a color performed by a video card.

7. The method of claim 5, wherein the output digital data represents a coordinate performed by a video card.

8. The method of claim 5, wherein the first number has r bits, and the second number has (m−r) bits.

9. The method of claim 5, wherein the different value is equal to a value of the second intermediate digital data minus a value of the input digital data.

10. The method of claim 9, wherein the rule is:

when the difference value is larger than or equal to 0.5*(2^m−1), adding one to the first number of the first intermediate digital data to obtain the output digital data;

when the difference value is less than −0.5*(2^m−1), subtracting one from the first number of the first intermediate digital data to obtain the output digital data; and

when the difference value is not larger than or not equal to 0.5*(2^m−1), and not less than −0.5*(2^m−1), the first intermediate digital data is equal to the output digital data.

11. A method for converting a first video data having m bits into a second video data having n bits, wherein n>m, the method comprising:

duplicating the first video data by bit order to form a first intermediate video data having n bits;

swapping r most significant bits and the remaining (m−r) bits of the first video data to obtain a second intermediate video data, wherein r=n−m*q, 0≦r<m;

obtaining a difference value between a value of the second intermediate video data and a value of the first video data; and

modifying the r least significant bits in the first intermediate video data by a rule referring the difference value to obtain the second video data.

12. The method of claim 11, wherein the second video data represents a color.

13. The method of claim 11, wherein the second video data represents a coordinate.

14. The method of claim 11, wherein the rule is:

when the difference value is larger than or equal to 0.5*(2^m−1), adding one to a value of the r least significant bits in the first intermediate video data to obtain the second video data;

when the difference value is less than −0.5*(2^m−1), subtracting one from a value of the r least significant bits in the first intermediate video data to obtain the second video data; and

when the difference value is not larger than or not equal to 0.5*(2^m−1), and not less than −0.5*(2^m−1), the first intermediate video data is equal to the second video data.