Advanced Micro Devices Corporation

Smeeta Gupta

Brian D McMinn

Thomas W Lynch

Robert M Perlman

Rod G Fleck

William M Johnson

Philip M Freidin

Cheng-Gang Kong

Gigy Baror

Timothy A Olson

Ole H Moller

Brian W Case

David I Sorensen

A floating point processor for performing arithmetic operations on floating point numbers includes a first arithmetic operation unit configured to operate on normalized numbers and a second arithmetic operation unit which includes a denormalizer for denormalizing normalized numbers and a normalizer for normalizing denormalized numbers. Each arithmetic operation unit has first and second inputs for receiving first and second operands, respectively, and an output for transmitting a result of the arithmetic operation. When a denormalized operand is presented as an input to the arithmetic operation unit configured to operate on normalized numbers, the denormalized input operand is redirected through the second arithmetic unit for normalization of the denormalized operand. The first arithmetic operation unit then performs its arithmetic operation using the normalized input operands. The result of the arithmetic operation is then analyzed to determine whether it has a zero or negative exponent. If the result has a zero or negative exponent, the result is directed through the second arithmetic unit a second time so that the result is denormalized. The denormalized result is then output.

Normalizing pipelined floating point processing unit

A floating point processor for performing arithmetic operations on floating point numbers includes a first arithmetic operation unit configured to operate on normalized numbers and a second arithmetic operation unit which includes a denormalizer for denormalizing normalized numbers and a normalizer for normalizing denormalized numbers. Each arithmetic operation unit has first and second inputs for receiving first and second operands, respectively, and an output for transmitting a result of the arithmetic operation. When an denormalized operand is presented as an input to the arithmetic operation unit configured to operate on normalized numbers, the denormalized input operand is redirected through the second arithmetic unit for normalization of the denormalized operand. The first arithmetic operation unit then performs its arithmetic operation using the normalized input operands. The result of the arithmetic operation is then analyzed to determine whether it has a zero or negative exponent. If the result has a zero or negative exponent, the result is directed through the second arithmetic unit a second time so that the result is denormalized. The denormalized result is then output.

A streamlined instruction processor processes data in response to a program composed of prespecified instructions in pipeline cycles. The processor comprises an instruction fetch unit, including an instruction interface adapted for connection to an instruction memory and for fetching instructions from the instruction memory. The instruction fetch unit includes an instruction prefetch buffer coupled to the instruction interface for buffering a sequence of instructions supplied to the instruction interface. A branch target cache is coupled with the prefetch buffer for storing sets of instructions retrieved from a corresponding set of locations in the instruction memory, having sequential instruction addresses. The first instruction in each such set is a branch target instruction in the program. In addition, an execution unit including a data interface adapted for connection to the data memory, executes the instructions in pipeline cycles. The execution unit includes a storage facility, coupled to the data interface, for storing data in a file of data locations identified by file addresses. The storage facility includes at least two read ports and one write port operable in response to file addresses. An addressing unit coupled to receive the instructions from the instruction register, supplies the file addresses to the read ports and the write port under program control. In addition, the addressing unit is operable in response to a stack pointer providing dynamic allocation of the file of data locations to processes within the program. A memory management unit is coupled to the data interface. The memory management unit includes an address interface adapted for connection to the data memory and the instruction memory for supplying instruction addresses to the instruction memory and data addresses to the data memory, in a simple single access mode, a pipeline mode and a burst mode.

Streamlined instruction processor

Growing community of inventors

Saratoga, CA, United States of America

Smeeta Gupta