Floating-point accumulator

• US 6,073,155 A
• Filed: 07/28/1997
• Issued: 06/06/2000
• Est. Priority Date: 08/08/1996
• Status: Expired due to Term

• Current Assignee: Fuji Xerox Company Limited (Xerox Holdings Corp.)
• Original Assignee: Fuji Xerox Company Limited (Xerox Holdings Corp.)
• Inventors: Toyoda, Shinjiro, Miyakawa, Nobuaki, Inabata, Shinjiro, Yamada, So
• Primary Examiner(s): Mai, Tan V.
• Application Number:
  US08/901,671
• Time in Patent Office: 1,044 Days
• Field of Search: 364/736.01-736.03, 364/750.5, 364/748.01, 364/748.11, 708/490, 708/495
• US Class Current: 708/490
• CPC Class Codes: G06F 7/485 Adding; Subtracting G06F7/4833, G06F7/4836 take precedence

Abstract

To obtain the sufficiently precise result of floating-point accumulation even if the quantity of computation is enormous, a floating-point accumulator according to the present invention is constituted as follows:

When two floating-point data are stored in any of shift registers, the two data are respectively output to BUS0 and BUS1 via one connected to the shift register of buffers. The two output data are input to an adder via BUS0 and BUS1 and output as added result data after adding the floating-point numbers. The above added result data is returned to each input of the shift registers via BUSW and a multiplexer and written into the shift register corresponding to the addition of the higher level by one of the shift register holding floating-point data before addition. The floating-point numbers are accumulated by repeating the above operation.

Claims

1. A floating-point accumulator for accumulating data represented by a floating point number, comprising:

input means for inputting input data to be accumulated;

adding means for adding a first pair of data and outputting the added first pair of data as added result data; and

data added order changing means for receiving the input data to be accumulated from the input means and the added result data, obtaining a second pair of data at a same addition level from at least one of the input data to be accumulated and the added result data, and inputting the second pair of data to the adding means as the first pair of data.

2. A floating-point accumulator according to claim 1, wherein the data added order changing means comprises:

an input data extracting section for extracting a third pair of data from the input data to be accumulated;

an input data storing section for storing the third pair of data extracted by the input data extracting section;

an added result obtaining section for obtaining a fourth pair of data from the added result data;

plural added result storing sections for storing the fourth pair of data obtained by the added result obtaining section for every addition level; and

a stored data retrieving section for retrieving the second pair of data from at least one of the input data storing section and the plural added result storing sections and inputting the second pair of data to the adding means.

3. A floating-point accumulator according to claim 2, wherein if at least one piece of data of the second pair of data does not exist, at least one piece of data of the second pair of data uses data `0` in place of the data which does not exist.

4. A floating-point accumulator according to claim 2, wherein the stored data retrieving section retrieves the second pair of data, giving priority to addition at a lower addition level.