Fujitsu Corporation

Fujitsu Microelectronics Limited

Fujitsu Semiconductor Limited

Seiji Suetake

Hideyuki Iino

Koichi Hatta

Tatsuya Nagasawa

Shunsuke Kamijo

Satoshi Matsui

Syuji Nishida

Kenji Furuya

Koji Takenouchi

Koichi Kuroiwa

Hiroyuki Fujiyama

Shinichi Utsunomiya

Noriko Kadomaru

Yukihiro Ozawa

Kenji Shirasawa

Makoto Miyagawa

A direct memory access controller is provided, in which an internal storage section storing control setting information; and a control section loading the control setting information from an external storage section to the internal storage section when a transfer request signal does not belong to a first group, and not loading the control setting information from the external storage section to the internal storage section when the transfer request signal belongs to the first group; are included, and a data transfer by a direct memory access is performed in accordance with the control setting information within the internal storage section.

Direct memory access controller

A system for DMA transfer includes a DMA controller, a bus connected to the DMA controller, a bus interface connected to the bus, and a plurality of registers coupled to the bus via the bus interface, wherein the bus interface is configured to allocate the plurality of registers doubly to nonconsecutive addresses and consecutive addresses to allow the DMA controller to access the plurality of registers through the consecutive addresses.

System and method for DMA transfer

A microcomputer that can increase the usage efficiency of a cache memory and increase the process speed is provided. In this microcomputer, a group of registers hold cache usage information that specifies whether the cache memory is to be used in execution of a process. When processes to be executed are switched, a process switch control circuit obtains the cache usage information of the next process from the group of registers, and stores the cache usage information in a first register. After the storing of the cache usage information in the first register, a cache control circuit stores the cache usage information in a second register. In accordance with the cache usage information stored in the second register, the cache control circuit puts the cache memory in a usable state or an unusable state.

Microcomputer, method of controlling cache memory, and method of controlling clock

A data controlling unit activates a predetermined number of data terminals according to a mode signal and changes a bus width of external data signal. According to the mode signal, an address controlling unit selects a predetermined number of bits of an internal address signal outputted from a controller and outputs the selected bits as an external address signal. Specifically, the address controlling unit selects upper bits of the internal address signal when the bus width of the external data signal is increased according to the mode signal. Therefore, it is possible to prevent occurrence of an unused external address terminal, enabling the increase in accessible external memory capacity.

Semiconductor integrated circuit having changeable bus width of external data signal

A circuit for changing clocks includes a clock generating circuit which generates an output clock signal by controlling a frequency of an original clock signal, and a control circuit which controls the clock generating circuit in response to an operation mode change signal indicative of a change from a first operation mode to a second operation mode of an external circuit operating based on the output clock signal, thereby changing the output clock signal from a first frequency corresponding to the first operation mode to an intervening frequency and then from the intervening frequency to a second frequency corresponding to the second operation mode, the intervening frequency having a frequency between the first frequency and the third frequency.

Clock shift circuit for gradual frequency change

A central processing system can maintain an efficient information reading operation even when a program executed by a central processing unit contains many branch commands. A prefetch queue of the central processing unit reads and information expected to be processed next by the central processing unit from a main memory. The function of the prefetch queue is deactivated in accordance with a control signal provided from a prefetch queue control unit. A block transfer function of a cache memory is also deactivated when unnecessary information is read from the main memory in accordance with the block transfer function.

Central processing unit and system having a prefetch queue and a command cache to perform an efficient information reading operation

The central processing unit is provided with an instruction queue storage section. This central processing unit is made of a memory, such as FIFO memory, that adopts first-in first-out method. A counter counters each time an instruction datum is stored in the instruction queue storage section. When the value of the counter is 0 or 1 and instruction fetch is not suppressed, a fetch request is issued.

Central processing unit and system counting instructions in a queue storage and outputting a fetch request when the count value is 0 or 1 to produce a next instruction address output

An information processing device includes a central processing unit, a cache memory unit and first and second decision circuits. The first decision circuit identifies one of partitioned address areas to be accessed before the central processing unit accesses the cache memory unit. The second decision circuit determines whether the above one of the partitioned address areas is a cachable area or a non-cachable area before address tag data is referred to in the cache memory unit.

Partitioned cache memory with switchable access paths

Information processing device with decision circuits and partitioned

An arithmetic operation unit for operating according to pipeline control and an instruction decoder for controlling the arithmetic operation unit by decoding an instruction, including a state retaining unit for retaining a state of the operation of the arithmetic operation unit, wherein the instruction decoder controls the execution of the arithmetic operation unit according to the information stored by the state retaining unit. A state is set when the decoder issues a signal for starting the arithmetic operation unit and the state is cleared when the decoder issues a signal for stopping the operation of the arithmetic operation unit. The arithmetic operation unit further comprises a unit for obtaining a maximum and a minimum value with a simple construction. A multiplier of the arithmetic operation unit comprises a unit for performing an addition of an exponential part of a multiplier and that of a multiplicand with a simple construction. The arithmetic operation unit further comprises a data packing and unpacking unit for packing and unpacking vector data. The data processing device is divided into several units and scan paths are divided into several paths corresponding to respective units.

Pipelined data processing device having improved hardware control over

A processor comprises a command storage unit for storing a plurality of commands and data received from outside the processor, a command interpreter for interpreting commands and data stored in the command storage unit, an address designator for designating a particular execution address of the storage unit according to a command interpreted by the command interpreter or to an operation start command, and an update selector for selecting whether or not to update the value of an execution address designated by the address designator according to a command interpreted by the command interpreter.

Coprocessor for performing an arithmetic operation by automatically

A slave processor for executing for example a vector operation is connected to a master processor. A vector length for a vector operation set to the slave processor can be changed without intervention of the master processor. When the master processor activates the slave processor, the slave processor outputs a busy signal immediately (at most one cycle later). The master processor reads the value of a busy register representing a busy/ready status of the slave processor in a slave access cycle at highest speed (in two cycles at most). Regardless of whether the master processor and the slave processor was designed as series products or general purpose products, they can be effectively connected.

Growing community of inventors

Kawasaki, Japan

Seiji Suetake