Circuit design method, apparatus, and program
IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0706238
(2003-11-12)
|
우선권정보 |
JP-P2002-331677(2002-11-15) |
발명자
/ 주소 |
|
출원인 / 주소 |
|
대리인 / 주소 |
|
인용정보 |
피인용 횟수 :
0 인용 특허 :
1 |
초록
▼
A circuit design method able to design a processing circuit to be small in scale when designing a processing circuit for performing a plurality of different processings on predetermined data, comprising a first step of identifying second processings performing the same processing on the same data am
A circuit design method able to design a processing circuit to be small in scale when designing a processing circuit for performing a plurality of different processings on predetermined data, comprising a first step of identifying second processings performing the same processing on the same data among pluralities of second processings forming each of a plurality of first processings when designing a processing circuit for applying a plurality of different first processings on predetermined data and a second step of designing a processing circuit having a processing circuit shared by the plurality of first processings and for performing the second processings identified at the first step.
대표청구항
▼
The invention claimed is: 1. A circuit design method executed by a computer for designing a processing circuit for applying a plurality of different first processings to predetermined data comprising: a first step of identifying second processings performing the same processing on the same data in
The invention claimed is: 1. A circuit design method executed by a computer for designing a processing circuit for applying a plurality of different first processings to predetermined data comprising: a first step of identifying second processings performing the same processing on the same data in pluralities of second processings forming each of said plurality of different first processings; a second step of designing a processing circuit comprising a first processing circuit shared by said plurality of different first processings and performing said second processings identified in said first step, and a second processing circuit for performing processings other than said second processings identified in said first step in said pluralities of second processings forming each of said plurality of different first processings; and when said plurality of different first processings are processings applying first linear transforms to said predetermined data different predetermined number of times, a third step of defining a second linear transform by combining a number of first linear transforms corresponding to a predetermined number of times of processing for each of the plurality of different first processings, wherein said first step further comprises identifying said second processings performing the same processing on the same data among said plurality of second processings forming said second linear transform defined for each of said plurality of different first processings at said third step. 2. The circuit design method as set forth in claim 1, further comprising, in said second step, designing said processing circuit so as to perform said plurality of different first processings in parallel on said predetermined data based on said second linear transforms defined in said third step. 3. The circuit design method as set forth in claim 1, wherein: said predetermined data is expressed by a vector by a predetermined base on a predetermined linear space, and said predetermined linear transforms are transforms defined on said linear space. 4. The circuit design method as set forth in claim 3, further comprising, when said predetermined linear space is indicated by the following (3-1), data "a" of the predetermined data is indicated as an m-dimensional vector by the following (3-4) when using the base shown in the following (3-2) as the predetermined base and said data "a" is indicated as in the following (3-3), said first linear transform is defined as a linear transform D on the linear space shown in the following (3-1), data "b" of the result of the above plurality of processings is shown as a k-dimensional vector by the following (3-5), and data bi indicating the results of the processings forming the data "b" shown in the following (3-5) is shown as a di-dimensional vector by the following (3-6), defining matrix M comprised by di rows and "m" columns, performing said second linear transforms, and shown by the following (3-7) in said third step and identifying said second processings performing the same processing on the same data among said plurality of second processings based on the following (3-7) defined in said third step, where, "m" and di are integers of 2 or more, the predetermined number of times corresponding to at least one of the above plurality of processings is 2 or more, and "k" is an integer of 2 or more: 5. The circuit design method as set forth in claim 4, wherein when using the base shown by the following (3-8) as said predetermined base and said data "a" is shown as in the following (3-9), said data "a" is shown by the following (3-10) as an m-dimensional vector: 6. The circuit design method as set forth in claim 4, wherein said third step defines said matrix M comprised of said matrixes D for performing γr-times multiprocessing based on the dimension γ on said linear space.
이 특허에 인용된 특허 (1)
-
Dawes Robert L. (Allen TX), Adaptive processing system having an array of individually configurable processing components.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.