JPH0417026A - Data clipping device - Google Patents

Abstract

PURPOSE:To decide an overflow and to make high speed clipping available with simple circuit configuration by comparing code bits before and after an arithmetic operation with each other. CONSTITUTION:An arithmetic means 300 which inputs input data including code bits and a first means 100 which compares a code bit obtained from the calculated result of the arithmetic means 300 with the code bit of the input data and discriminates the occurrence of a positive or negative overflow are provided. In addition, a second means 200 which clips the calculated result to the positive maximum value when the occurrence of an positive overflow is discriminated and to the negative minimum value when the occurrence of an negative overflow is discriminated. Therefore, high-speed clipping can be realized with simple comparison.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data clipping device when an overflow occurs in an arithmetic operation in digital signal processing.

[Prior Art] Conventionally, regarding data clipping after arithmetic operations, the obtained result is compared with a predetermined maximum positive value and minimum negative value, and if overflow occurs, the maximum or minimum value is set. Use the clip method.

[Problems to be Solved by the Invention] In the conventional example described above, it is necessary to compare all bits of the result of the arithmetic operation with the maximum value and the minimum value. For this reason, there is a problem in that the comparison means becomes complicated and extra time is required for the overall calculation.

An object of the present invention is to provide a clipping device that enables high-speed clipping through simple comparison.

[Means for Solving the Problems] The present invention includes a calculation means for inputting input data including a sign bit, and a comparison between the sign bit obtained from the calculation result of the calculation means and the sign bit at the time of the input data, Both are O
When , it is determined that a correct result with a positive value has been obtained, and both of them are 1.
When , it is determined that a correct negative value result has been obtained, and when the sign bit of the input data is 0 and the sign bit obtained from the operation result is 1, it is determined that a positive overflow has occurred.
The first means determines that a negative overflow has occurred when the sign bit of the input data is 1 and the sign bit obtained from the operation result is 0, and the first means is obtained from the operation result when it is determined that the above positive overflow has occurred. Replace the sign bit with O, set all data bits to 1, clip to the maximum positive value, and when it is determined that the above negative overflow has occurred, replace the sign bit obtained from the operation result with 1, The method further comprises: a second means for setting all data bits to 0 and clipping them to a minimum negative value (claim 1).

Furthermore, the present invention provides an arithmetic means that inputs input data with duplicated sign bits, uses one sign bit as an overflow reflection bit due to an operation, and outputs the other sign bit as is, and an arithmetic operation of the arithmetic means. Compare the sign bit after the operation obtained from the result with the sign bit that is output as is, and when both are O, it is determined that a positive value is correct, and when both are 1, a negative value is correct. It is determined that a result has been obtained, and when the sign bit that is output as is is O and the sign bit after the operation is 1, it is determined that a positive overflow has occurred, and the sign bit that is output as is is 1 and the sign after the operation is The first means determines that a negative overflow has occurred when the bit is 0, and the first means replaces the sign bit after the operation with 0 and sets all data bits to 1 when it is determined that a positive overflow has occurred. When it is determined that the above negative overflow has occurred, the sign bit after the operation is replaced with 1, and all data bits are set to 0 to clip to the maximum negative value. (Claim 2)

[Operation] According to the present invention, an overflow can be determined simply by comparing the sign bits before and after an operation, and clipping is performed when this overflow occurs (claim 1).

Further, according to the present invention, since the sign bits before and after the operation are obtained through the calculation means, overflow can be determined online by looking only at the output of the calculation means (Claim 2).
).

〔Example〕

FIG. 1 is a diagram showing an embodiment of the clip device of the present invention. In this embodiment, the arithmetic (-both manual and multi-input) calculation means 3
00, overflow detection means 100, clipping means 2
Consists of 00. These means may be replaced by units. Further, it may be replaced with a mechanism that processes software (including microprograms).

Arithmetic in the arithmetic operation means 300 is a broad concept that includes not only addition, subtraction, multiplication, and division, but also forward signal processing such as discrete cosine transformation in the case of negative input, and the common point between the two is the occurrence of overflow. This is a possibility.

The arithmetic operation means 300 receives input data (S, D, . . . DI) 12 including a sign bit S1. S
l is MSB (most significant bit), D is LSB (least significant bit), sign bit S is positive at 51-0, S
Let l −1 represent negative.

In this embodiment, in addition to the code bit S1, the line 12 carries the second code bit S, which is the code bit itself.
is provided, and this is input to the calculation means 300 together with the input data. That is, the code bits are doubled.

On the other hand, in the arithmetic means 300 that receives the duplicated code bits Sl and Sz, DI
-Dn as well as the original operation target, and the sign bit S
z is treated as a signal to be directly output without being subjected to this original calculation. Therefore, if an overflow occurs as a result of the calculation, only the sign of Sl will change, and the sign of Sz will remain unchanged. The output of Sz is shown as S20, but S 2 = 3
It is 211.

The overflow detection means 100 compares the two code bits and outputs determination signals C+ and Cz. A specific circuit configuration of this detection means 100 is shown in FIG.

The logical relationship is shown in a portion of FIG.

The clipping means 200 is the overflow detection means 100
The overflow determination signals C, , C2 are input, and a predetermined clipping operation is performed based on the logical relationship between the determination signals C1 and C2 as shown in FIG. A specific circuit configuration is shown in FIG.

In FIG. 1, S, ID, . . . Dll represent data including a sign bit after operation, and Sl. D. one...
...DI2 is the output data of the clipping means. Furthermore,
Sz. is the output sign bit of the calculation means 300 corresponding to the sign bit S2.

Furthermore, since there are various specific circuit configurations of the arithmetic operation means 300, the drawings are omitted.

Now, in FIG. 2, the overflow detection means 100 consists of two AND gates IOL 102 which receive two inputs S ll+ 320 exclusively. Sign bit Sl
and Sol are both "1" or both are 0", C1-0
2-0, if the sign bits Sll and S20 do not match, one of C, C2 outputs 1 and the other outputs O.

In FIG. 3, the clip means 200 includes n+1 switches 20.
1 to 203. n+1 is the data bit length (n
) and the code bit length (1), and a switch is provided for each bit.

Each switch 201 to 203 has a judgment signal C, , C, GND indicating a logical "O", and ■C indicating a logical "1".
C1 and calculation outputs S, D□1 to DI+ are respectively input terminals 11. Enter IZ, 13.

The logical relationships among the switches 201-203 are shown in FIG.

FIG. 4 has the following relationships.

(a) When the sign bit Szo of the input data is O (that is, positive input data).

If the sign bit S11 after the operation is 0, it can be assumed that overflow will not occur, and in this case, C=C,=O,
The input data (SzD-+...Dl,) is output as is.

If S++ is 1, it can be considered that a positive overflow has occurred, and in this case, C1-1, CZ=O
Then, S1□ is replaced with 0, and D7□ to DI2 are all clipped to 1 and output. This is a clip to the maximum value. To clip to all 1, select CC.

(■), when the sign bit S2° of the input data is l (that is, negative input data).

If the sign bit S++ after the operation is O, it can be considered that a negative overflow has occurred, and in this case, C, -〇, CZ=
1, replaces S, □ with l, and clips and outputs D, 2 to DIZ to all 0s. This is a clip to the minimum value. For clipping to all 0s, select GND.

If the sign bit s++ after the operation is 1, it can be considered to be a correct negative value. In this case, C+ -C2=0, and the input data (S++, D, , +...Dl
, ) is output as is.

According to the above-described embodiment, overflow checking can be performed in an online state without storing the sign bit of input data.

In this embodiment, passing the sign bit as it is within the arithmetic means means not only passing it through, but also outputting it through various latch circuits used during the arithmetic process in the arithmetic means. .

In addition to the calculating means, other passing means such as gates may be used to
You may also obtain

〔Effect of the invention〕

According to the present invention, it is possible to check for overflow in an online state without storing the sign bit of input data, and high-speed ripping is possible with a simple circuit configuration.

[Brief explanation of drawings]

Fig. 1 is an embodiment of the clipping device of the present invention, Fig. 2 is an embodiment of the detection means, Fig. 3 is an embodiment of the clipping means, and Fig. 4 is the logical relationship between the detection means and the clipping means. FIG. 100... Overflow detection means, 200... Clipping means, 300... Calculating means, 101.1 (12.
...And gate, 201-203...switcher.

Claims (1)

[Claims] 1. An arithmetic means that inputs input data including a sign bit, and a sign bit obtained from the arithmetic result of the arithmetic means and a sign bit at the time of the input data, and when both are 0. When both are 1, it is determined that a correct result with a negative value has been obtained, and when the sign bit of the input data is 0 and the sign bit obtained from the operation result is 1. A first means of determining that a positive overflow has occurred when the input data is 1, and determining that a negative overflow has occurred when the sign bit of the input data is 1 and the sign bit obtained from the operation result is 0; When it is determined that a
a second means for clipping to the minimum negative value by replacing the sign bit obtained from the operation result with 1 and setting all data bits to 0 when it is determined that the negative overflow has occurred; Device 2, an arithmetic means that inputs input data with duplicated sign bits, uses one sign bit as an overflow reflection bit by the operation, and outputs the other sign bit as is; and a calculation result of the arithmetic means. Compare the obtained sign bit after the operation with the sign bit that is output as is, and when both are 0, it is determined that a positive result has been obtained, and when both are 1, a negative value is correct. If the sign bit that is output as is is 0 and the sign bit after the operation is 1, it is determined that a positive overflow has occurred, and the sign bit that is output as is is 1 and the sign bit after the operation is 1. The first means determines that a negative overflow has occurred when the value is 0, and when it is determined that the positive overflow has occurred, the sign bit after the operation is replaced with 0 and all data bits are set to 1 to generate a positive value. A second means of clipping to the maximum value of negative values, replacing the sign bit after the operation with 1, and setting all data bits to 0 when it is determined that the above-mentioned negative overflow has occurred. and a clip device consisting of.