Abstract:
Techniques for a massively parallel and memory centric computing system. The system has a plurality of processing units operably coupled to each other through one or more communication channels. Each of the plurality of processing units has an ISMn interface device. Each of the plurality of ISMn interface devices is coupled to an ISMe endpoint connected to each of the processing units. The system has a plurality of DRAM or Flash memories configured in a disaggregated architecture and one or more switch nodes operably coupling the plurality of DRAM or Flash memories in the disaggregated architecture. The system has a plurality of high speed optical cables configured to communicate at a transmission rate of 100 G or greater to facilitate communication from any one of the plurality of processing units to any one of the plurality of DRAM or Flash memories.
Abstract:
Techniques for operating a DIMM apparatus. The apparatus comprises a plurality of DRAM devices numbered from 0 through N−1, where N is an integer greater than seven (7), each of the DRAM devices is configured in a substrate module; a buffer integrated circuit device comprising a plurality of data buffers (DB) numbered from 0 through N−1, where N is an integer greater than seven (7), each of the data buffers corresponds to one of the DRAM devices; and a plurality of error correcting modules (“ECMs”) associated with the plurality of DRAM devices.
Abstract:
The present systems include a memory module containing a plurality of RAM chips, typically DRAM, and a memory buffer arranged to buffer data between the DRAM and a host controller. The memory buffer includes an error detection and correction circuit arranged to ensure the integrity of the stored data words. One way in which this may be accomplished is by computing parity bits for each data word and storing them in parallel with each data word. The error detection and correction circuit can be arranged to detect and correct single errors, or multi-errors if the host controller includes its own error detection and correction circuit. Alternatively, the locations of faulty storage cells can be determined and stored in an address match table, which is then used to control multiplexers that direct data around the faulty cells, to redundant DRAM chips in one embodiment or to embedded SRAM in another.
Abstract:
A DRAM refresh method used with a memory system organized into rows of memory cells, each of which has an associated data retention time, with the system arranged to refresh predefined blocks of memory cells simultaneously. For each block of memory cells that are to be refreshed simultaneously, the minimum data retention time for the memory cells in the block is determined. Then, an asymmetric refresh sequence is created which specifies the order in which the blocks of memory cells are refreshed, such that the blocks having the shortest minimum data retention times are refreshed more often than the blocks having longer minimum data retention times.
Abstract:
A random access memory includes a plurality of memories configured to store and provide data, and a test module coupled to the plurality of memories, wherein the test module is configured to write a first write data pattern into at last a first portion of the plurality of memories in response to a data pattern value, wherein the test module is configured to read a read data pattern from the plurality of memories, wherein the test module is configured to compare the first write data pattern to the read data pattern, and wherein the test module is configured to report errors in response to a comparison of the write data pattern to the read data pattern.
Abstract:
In an example, the present invention provides a computing system. The system has a memory interface device comprising a counter, a dynamic random access memory device coupled to the memory interface device. The device comprises a plurality of banks, each of the banks having a subarray, each subarray having a plurality of memory cells. The device has a data interface coupled to the plurality of banks. The device has an address interface coupled to the plurality of banks, and a particular pre-charge command configured to be transferred to the memory interface device. The counter is adapted to count a measured time duration from a first time when data are available at the data interface to a second time when a pre-charge command is received by the address interface.
Abstract:
A buffer integrated circuit device. The device comprising an output driver formed on the substrate member, the output driver having at least a command bus and an address bus. The device has a protocol and parity checking block (“Block”). The device has a table configured in the block. The table is programmable with a plurality of timing parameters. The device has a memory state block coupled to the table and a command history table coupled to the table to process protocol information for all commands that pass through the Block. The buffer integrated circuit device utilizes the protocol checking functionality to prevent failure propagation and enables data protection even in the case of host memory controller failure or system-level failure of any signal or signals on the command, control and address bus from the host memory controller to the buffer integrated device.
Abstract:
An interface device for a memory module comprising a plurality of DRAMs includes a memory configured to store DRAM test program instructions, and a programmable processing device coupled to the memory, wherein the programmable processing device is configured to receive input data and input memory addresses from an external processor, wherein the programmable processing device is configured to provide data and memory addresses to the plurality of DRAMs, and wherein the programmable processing device is programmed to perform operations specified by the DRAM test program instructions.
Abstract:
A method for operating a DRAM device. The method includes receiving in a memory buffer in a first memory module hosted by a computing system, a request for data stored in RAM of the first memory module from a host controller of the computing system. The method includes receiving with the memory buffer, the data associated with a RAM, in response to the request and formatting with the memory buffer, the data into a scrambled data in response to a pseudo-random process. The method includes initiating with the memory buffer, transfer of the scrambled data into an interface device.
Abstract:
A hybrid memory system. This system can include a processor coupled to a hybrid memory buffer (HMB) that is coupled to a plurality of DRAM and a plurality of Flash memory modules. The HMB module can include a Memory Storage Controller (MSC) module and a Near-Memory-Processing (NMP) module coupled by a SerDes (Serializer/Deserializer) interface. This system can utilize a hybrid (mixed-memory type) memory system architecture suitable for supporting low-latency DRAM devices and low-cost NAND flash devices within the same memory sub-system for an industry-standard computer system.