Abstract:
A bonded assembly includes a memory die containing a memory device and a plurality of bit lines, and logic die bonded to the memory die. The logic die contains a control circuit configured to control operation of the memory device. The control circuit contains a peripheral circuit region, a sense amplifier region, and a power and control signal region located adjacent to the sense amplifier region and containing at least one power and control signal interconnect structure which is configured to provide a power or control signal to or from the peripheral circuit region.
Abstract:
Devices and methods related to spiking neural units in memory. One device includes a memory array and a complementary metal-oxide semiconductor (CMOS) coupled to the memory array and located under the memory array, wherein the CMOS includes a spiking neural unit comprising logic configured to receive an input to increase a weight stored in a memory cell of the memory array, collect the weight from the memory cell of the memory array, accumulate the weight with an increase based on the input, compare the accumulated weight to a threshold weight, and provide an output in response to the accumulated weight being greater than the threshold weight.
Abstract:
The embodiments described herein describe technologies for non-volatile memory persistence in a multi-tiered memory system including two or more memory technologies for volatile memory and non-volatile memory.
Abstract:
Examples are disclosed for accessing a dynamic random access memory (DRAM) array. In some examples, sub-arrays of a DRAM bank may be capable of opening multiple pages responsive to a same column address strobe. In other examples, sub-arrays of a DRAM bank may be arranged such that input/output (IO) bits may be routed in a serialized manner over an IO wire. For these other examples, the IO wire may pass through a DRAM die including the DRAM bank and/or may couple to a memory channel or bus outside of the DRAM die. Other examples are described and claimed.
Abstract:
According to an example, a method for adaptive-granularity row buffer (AG-RB) caching may include determining whether to cache data to a RB cache, and adjusting, by a processor or a memory side logic, an amount of the data to cache to the RB cache for different memory accesses, such as dynamic random-access memory (DRAM) accesses. According to another example, an AG-RB cache apparatus may include a 3D stacked DRAM including a plurality of DRAM dies including one or more DRAM banks, and a logic die including a RB cache. The AG-RB cache apparatus may further include a processor die including a memory controller including a predictor module to determine whether to cache data to the RB cache, and to adjust an amount of the data to cache to the RB cache for different DRAM accesses.
Abstract:
Techniques and mechanisms for determining a timing of a command to access a memory device resource. In an embodiment, a multi-cycle command which is exchanged from a memory controller to a memory device, wherein the multi-cycle command indicates an access to a bank of the memory device. Timing of the one or more other commands is controlled, based on the multi-cycle command, to enforce a time delay parameter which describes an operational constraint of the memory device. In another embodiment, timing of one or more commands is determined with reference to a beginning of a last cycle of a multi-cycle command.
Abstract:
An access of data in a memory device is sampled. In response to the sampled access of data, a refresh operation is performed in the memory device.
Abstract:
A detection circuit is provided for a particular group of memory cells in a memory device, where the detection circuit is to be updated in response to at least one access of data and at least one neighboring group of memory cells. The particular group of memory cells is refreshed in response to an indication from the detection circuit, where the indication indicates presence of potential disturbance of the particular group of memory cells.
Abstract:
Systems and methods are disclosed that may include a first layer comprising a first redundant memory element, an input/output interface, a first layer fuse box, and a fuse blowing control. These systems and methods also may include a second layer coupled to the first layer through a first connection comprising a second layer memory element and a second layer fuse box coupled to the first redundant memory element. In addition, these systems and methods may further include a redundancy register coupled to the first layer, wherein upon the failure of part of the second layer memory element, the redundancy register provides information to the fuse blowing control that allocates part of the first redundant memory element to provide redundancy for the failed part of the second layer memory element by blowing elements in the first layer fuse box and the second layer fuse box.