Electrical computers: arithmetic processing and calculating – Electrical digital calculating computer – Particular function performed
Reexamination Certificate
1998-07-07
2001-05-15
Malzahn, David H. (Department: 2121)
Electrical computers: arithmetic processing and calculating
Electrical digital calculating computer
Particular function performed
Reexamination Certificate
active
06233597
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a computing method and a computing apparatus which executes a double-precision multiplication by using a single-precision multiplying device.
2. Description of the Prior Art
Conventionally, a double-precision multiplication system is known in which products of the high-order word/low-order word of a double-precision multiplicand and the high-order word/low-order word of a double-precision multiplier are obtained by using a single-precision multiplying device and a digit place alignment addition operation is performed on each of the products, thereby obtaining a double-precision multiplication result. This system is disclosed in, for example, Japanese patent publication (Kokai) No. HEI8-30439. In the prior art example disclosed in Japanese patent publication (Kokai) No. HEI8-30439, means for holding the most significant bit of low-word of a double-precision multiplier is disposed, and the most significant bit of the low-order word of the double-precision multiplier is used in the encoding of the high-order word of the double-precision multiplier, so that the low-order word of the double-precision multiplier is treated as a signed binary. In order to perform a multiplication of the low-order word of the double-precision multiplicand, the system is provided with multiplying means having a function of enabling a multiplication even when the multiplicand is an unsigned binary. Therefore, the prior art technique can realize a double-precision multiplication without impairing the precision.
In the prior art example disclosed in Japanese patent publication (Kokai) No. HEI8-30439, however, a sign indicative of a positive number is assumed at the bit which is higher by one bit than the most significant bit of a multiplicand, when the multiplying means treats the multiplicand as an unsigned binary. Therefore, the digit place of the sign of the multiplication result is different from that of a result of a multiplication in which the multiplicand is a signed binary. Consequently, the shift number for digit place alignment is different in additions of products of the high-order word/low-order word of a double-precision multiplicand and the high-order word/low-order word of a double-precision multiplier, and hence the prior art example has a problem in that the circuit scale for digit place aligning means becomes large.
SUMMARY OF THE INVENTION
It is an object of the invention to provide a computing method and a computing apparatus which can solve the problem, in which the circuit scale for digit place aligning means can be reduced, and which can realize a double-precision multiplication with an excellent efficiency.
The computing method of the invention is characterized in that, in a binary fixed-point number system in which a most significant bit is a sign bit and a decimal point is between the most significant bit and a one-bit lower bit, when products of a high-order word/low-order word of a double-precision multiplicand and a high-order word/low-order word of a double-precision multiplier are to be obtained by using a single-precision multiplying device, and a digit place alignment addition operation is to be performed on the obtained products to obtain a double-precision multiplication result, at least two digits are set before the decimal point, thereby allowing each of the products of the high-order word/low-order word of the double-precision multiplicand and the high-order word/low-order word of the double-precision multiplier, to be obtained at a bit width which is larger by at least one bit than a bit width of double precision.
When at least two digits are set before a decimal point as described above, the shift number for digit place alignment of each of the products obtained at a bit width which is larger by at least one bit than a bit width of double precision is relatively restricted to the same bit width. Therefore, the circuit scale for digit place aligning means can be reduced, and a double-precision multiplication can be efficiently realized.
By setting at least three digits before a decimal point, each of the products of the high-order word/low-order word of a double-precision multiplicand and the high-order word/low-order word of a double-precision multiplier may be obtained at a bit width which is larger by two, three, or more bits than a bit width of double precision. This is nothing but sign extension of a product obtained at a bit width which is larger by one bit, and hence results only in an increased circuit scale. Usually, it is therefore sufficient to set two digits before a decimal point so as to obtain products at a bit width which is larger by one bit than a bit width of double precision. Hereinafter, the computing method of the invention will be described under assumption that products of the high-order word/low-order word of a double-precision multiplicand and the high-order word/low-order word of a double-precision multiplier are obtained at a bit width which is larger by one bit than a bit width of double precision.
In the computing method, when a product of the low-order word of the double-precision multiplicand and the low-order word of the double-precision multiplier is to be obtained, for example, a data in which a sign bit indicative of a positive number is added to a digit that is higher by one bit than the most significant bit of the low-order word of the double-precision multiplicand is multiplied with a data in which the most significant bit of the low-order word of the double-precision multiplier is a sign bit, a value “0” is added to a digit which is lower by one bit than the least significant bit of a result of the multiplication, and a result of the addition is output as a desired product.
When a product of the high-order word of the double-precision multiplicand and the low-order word of the double-precision multiplier is to be obtained, for example, a data in which a sign bit of the most significant bit of the high-order word of the double-precision multiplicand is subjected to one-bit sign extension is multiplied with a data in which the most significant bit of the low-order word of the double-precision multiplier is a sign bit, a value “0” is added to a digit which is lower by one bit than the least significant bit of a result of the multiplication, and a result of the addition is output as a desired product.
When a product of the low-order word of the double-precision multiplicand and the high-order word of the double-precision multiplier is to be obtained, for example, a data in which a sign bit indicative of a positive number is added to a digit that is higher by one bit than the most significant bit of the low-order word of the double-precision multiplicand is multiplied with a data in which a data having the same value as that of the most significant bit of the low-order word of the double-precision multiplicand is added to a digit of the least significant bit of the high-order word of the double-precision multiplier, a value “0” is added to a digit which is lower by one bit than the least significant bit of a result of the multiplication, and a result of the addition is output as a desired product.
When a product of the high-order word of the double-precision multiplicand and the high-order word of the double-precision multiplier is to be obtained, for example, a data in which a sign bit of the most significant bit of the high-order word of the double-precision multiplicand is subjected to one-bit sign extension is multiplied with a data in which a data having the same value as that of the most significant bit of the low-order word of the double-precision multiplier is added to a digit of the least significant bit of the high-order word of the double-precision multiplier, a value “0” is added to a digit which is lower by one bit than the least significant bit of a result of the multiplication, and a result of the addition is output as a desired product.
A single-precision multiplying device which is usually used performs a multiplication of two single-pre
Kabuo Hideyuki
Tanoue Kazufumi
Yamanaka Ryutaro
Malzahn David H.
Matsushita Electric - Industrial Co., Ltd.
Stevens Davis Miller & Mosher L.L.P.
LandOfFree
Computing apparatus for double-precision multiplication does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Computing apparatus for double-precision multiplication, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Computing apparatus for double-precision multiplication will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2533089