摘要:
Arrays such as SRAMs, DRAMs, CAMs & Programmable ROMs having multiple independent failures are repaired using redundant bit lines. A first embodiment provides redundant bit lines on one side of the array. During a write, data is shifted towards the redundant bit lines on the one side of the array, bypassing failed bit lines. A second embodiment provides a spare bit line on each side of the array. During a write, a first failing bit line is replaced by a first spare bit line on a first side of the array, and a second failing bit line is replaced by a second spare bit line on a second side of the array.
摘要:
Measurement methods and a ring oscillator circuit are provided for evaluating dynamic circuits. The ring oscillator circuit includes a one-shot pulse generator receiving a single transition input signal and producing a pulse output signal having a rising transition and falling transition. The dynamic circuit to be evaluated is coupled to an output of the one-shot pulse generator receiving the pulse output signal of the one-shot pulse generator and producing a delayed output pulse at an output. A divide-by-two circuit is coupled to the output of the dynamic circuit to be evaluated. An output signal of the divide-by-two circuit is fed back to the one-shot pulse generator, and the cycle is repeated, thus oscillating. A multiplexer is connected between output of the dynamic circuit to be evaluated and the divide-by-two circuit. The multiplexer receives the pulse output of the one-shot pulse generator and includes a select input for selecting the output of the dynamic circuit to be evaluated or the pulse output of the one-shot pulse generator. By inserting the evaluation circuit into a path that can be multiplexed in and out of the oscillator path, and by measuring the difference between the frequency with and without the evaluation circuit in the path, the performance of the evaluation circuit can be accurately determined.
摘要:
A method and apparatus for handling variable data word widths and array depths in an array built-in self-test system for testing a plurality of memory arrays using a single controller. Each array includes a predetermined row and column address depth and data word width. Each array further includes a scan register. A universal test data word is generated and sent to the scan register of each array. The universal length test data word has a length dependent upon the maximum row address depth, maximum column address depth and/or the maximum data word width. A portion of the test data word which exceeds the column address depth, row address depth and/or the data word width of a particular array is shifted off the end of the scan register of the particular array.
摘要:
A multi-threaded memory (and associated method) for use in a multi-threaded computer system in which plural threads are used with a single processor. The multi-threaded memory includes: multi-threaded storage cells; at least one write decoder supplying information to a selected multi-threaded storage cell; and at least one read decoder accessing information from a selected multi-threaded storage cell. Each of the multi-threaded storage cells includes: N storage elements, where N.gtoreq.2, each of the N storage elements having a thread-correspondent content; a write interface supplying information to the intra-cell storage elements; and a read interface reading information from the intra-cell storage elements. At least one of the intra-cell read and write interfaces selects one of the thread-correspondent contents based at least in part by identifying the corresponding thread to achieve intra-cell thread-correspondent content selection.
摘要:
An apparatus, program product, and method of testing a silicon-on-insulator (SOI) static random access memory (SRAM) introduce switching history effects to a memory cell during testing to stress the memory cell such that a reliable determination of stability may be made. Stress is applied to a memory cell through the use of a bitline precharge stress operation, which utilizes the bitline pairs coupled to a memory cell to attempt to flood the memory cell with charge and thereby attempt to cause the memory cell to unexpectedly switch state. The bitline precharge stress operation is performed immediately after the memory cell has been switched to one state after being maintained in an opposite state for a length of time that is sufficient to introduce switching history effects to the memory cell. While a bitline precharge operation may be implemented separate from any write operation, the bitline precharge stress operation may also be incorporated into a write operation through delaying the deassertion of the wordline that occurs in a conventional write operation until after initiation of the bitline precharge operation that conventionally occurs near the end of such a write operation.
摘要:
A method and apparatus are provided for laser fuseblow protection in transistors, such as silicon-on-insulator (SOI) transistors. The transistors are connected to a fuse. A pair of diodes are connected in series between a high supply and ground. A common connection of the series connected pair of diodes is connected to a common connection of the fuse and transistors. A charge is shunted to the high supply or ground by the pair of diodes with a first voltage a set value above the high supply and a second voltage a set value below the ground. A pair of protection diodes are provided on each side of the fuse with transistors. The transistors are either connected to one side of the fuse or to both sides of the fuse.
摘要:
A method, an apparatus, and a computer program are provided to reading indicia from an SRAM cell. A low value is generated on a write true line. A high value is generated on a continuous bit_line. The true node of the SRAM cell is evaluated through use of a floating voltage coupled to the true node of the SRAM cell. If the floating voltage stays substantially constant, the value read from the SRAM cell is a high. If the floating voltage is drained to ground, the value read from the SRAM cell is a low.
摘要:
A high performance domino static random access memory (SRAM) is provided. The domino SRAM includes a plurality of local cell groups. Each of the plurality of local cell groups includes a plurality of SRAM cells and a local true bitline coupled to each of the plurality of SRAM cells of each local cell group. A continuous complement bitline is coupled to each of the plurality of local cell groups and is coupled to each of the plurality of SRAM cells of each local cell group. For a write to the SRAM cell complement node, only driving the continuous complement bitline is required. The domino SRAM reduces the number of required wires and required transistors as compared to prior art domino SRAM and thus the area needed and power consumption are reduced for the domino SRAM.
摘要:
A method and apparatus are provided for laser fuseblow protection in transistors, such as silicon-on-insulator (SOI) transistors. The transistors are connected to a fuse. A pair of diodes are connected in series between a high supply and ground. A common connection of the series connected pair of diodes is connected to a common connection of the fuse and transistors. A charge is shunted to the high supply or ground by the pair of diodes with a first voltage a set value above the high supply and a second voltage a set value below the ground. A pair of protection diodes are provided on each side of the fuse with transistors. The transistors are either connected to one side of the fuse or to both sides of the fuse.
摘要:
A high performance, low cell stress, low-power silicon-on-insulator (SOI) complementary metal oxide semiconductor (CMOS) latch-type sensing method and apparatus are provided. A silicon-on-insulator (SOI) complementary metal oxide semiconductor (CMOS) latch-type sense amplifier includes a precharge circuit for charging complementary bit and data lines to a predefined precharge voltage during a precharge cycle. The precharge voltage is lower than a full rail voltage. The reduced bit and data line precharge voltage substantially reduces voltage stress applied to the access transistors in the RAM cells. A pre-amplifying mechanism produces an offset voltage between the complementary data lines before the. sense amplifier is set. The pre-amplifying mechanism includes a pre-amplifying FET that is substantially smaller than a sensing silicon-on-insulator (SOI) field effect transistor (FET) in the sense amplifier. The pre-amplifying mechanism aids offset voltage development before the sense amplifier is set. The full rail voltage is provided for the complementary data lines when the sense amplifier is set. The full rail voltage can be applied during the write mode.