Abstract: An object code for sequentially switching contexts of processing circuits arrayed in a matrix in a parallel operation apparatus is generated from a general source code descriptive of operation of the parallel operation apparatus. A DFG is generated from the source code descriptive of operation of the parallel operation apparatus according to limiting conditions, registered in advance, representing a physical structure, etc. of the parallel operation apparatus, and scheduled in a CDFG. An RTL description is generated from the CDFG, converting a finite-state machine into an object code and converting a data path into a net list. An object code of the processing circuits is generated in each context from the net list.

Abstract: A SRAM has a plurality of four-transistor/no-load memory cells each operating in a precharge mode for maintaining stored data based on off-leak currents of MOSFETs. The SRAM has a reference voltage generator, an impedance converter, a plurality of word line drivers, and a plurality of word lines each extending along a row of the memory cells.

Abstract: A semiconductor memory device includes a pair of data lines, a precharging and equalizing circuit, a setting circuit and a data write circuit. The precharging and equalizing circuit is provided between the data lines to equally precharge the data lines to a first voltage in response to a precharge and equalize signal. The setting circuit is provided between the data lines to set one of the precharged data lines to a second voltage in response to data signals. The second voltage is lower than the first voltage. Also, a data Is written to a memory cell based on the second voltage of the one precharged data line and the first voltage of the other precharged data line. The data write circuit supplies the data signals to the setting circuit based on the data.