Abstract: Methods and systems are disclosed for reducing power consumption by a system including a digital unit and an optical unit. Techniques disclosed comprise generating a workload signature of an incoming workload to be executed by the system. Based on the generated workload signature, techniques disclosed comprise matching the incoming workload with a profile of stored workload profiles. The workload profiles are generated by a trace capture unit. Based on the associated profile, a task submission transaction is sent to the optical unit of the system, representative of a request to execute the incoming workload by the optical unit.

Abstract: Multi-level cell memory management techniques are described. In one example, the memory controller is configured to control whether a single-level cell operation or a multi-level cell operation to be used using different mapping schemes. The single-level cell operation, for instance, is usable to store a data word using two states whereas the multi-level cell operation is usable to store the data word by also using an intermediate state. In order to store the data word using two states, the memory controller is configurable to separate the data word across two word lines in the physical memory. In an implementation, use of the different operations and corresponding mapping schemes by the memory controller alternates between adjacent word lines in physical memory.

Abstract: A method includes storing a function representing a set of data elements stored in a backing memory and, in response to a first memory read request for a first data element of the set of data elements, calculating a function result representing the first data element based on the function.

Abstract: Compacted addressing for transaction layer packets, including: determining, for a first epoch, one or more low entropy address bits in a plurality of first transaction layer packets; removing, from one or more memory addresses of one or more second transaction layer packets, the one or more low entropy address bits; and sending the one or more second transaction layer packets.

Abstract: A page swapping memory protection system tracks accesses to physical memory pages, such as in a table with each row storing a physical memory page address and a counter value. This counter value records the number of accesses (e.g., read access or write accesses) to the corresponding physical memory page. In response to one of the counters exceeding a threshold value, the corresponding physical memory page is swapped with another page in physical memory (e.g., a page the table indicates has a smallest number of accesses). According, unreliability in the physical memory introduced due to frequent accesses to a particular physical memory page is mitigated.

Abstract: Systems, apparatuses, and methods for scattering floating point values to heterogeneous memory devices are disclosed. An inference engine performs floating point calculations during pre-training and during post-training operations. A scatter unit stores the floating point number values in multiple memories with different error correction capabilities. A first portion of each floating point number value is stored in a first memory having a relatively high error correction capability, and a second portion of each floating point number value is stored in a second memory with a relatively low error correction capability. In one scenario, the first portion includes the sign and exponent fields, while the second portion includes the mantissa field. The resiliency of the inference engine to overcome small errors allows for convergence to the final result in spite of any errors in the retrieved second portion.

Abstract: One or both of read and write accesses to a fabric-attached memory module via a fabric interconnect are monitored. In one or more implementations, offloading of one or more tasks accessing the fabric-attached memory module to a processor of a routing system associated with the fabric-attached memory module is initiated based on the read and write accesses to the fabric-attached memory module. Additionally or alternatively, replicating memory of the fabric-attached memory module to a cache memory of a computing node in the disaggregated memory system executing one or more tasks of a host application is initiated based on the write accesses to the fabric-attached memory module.

Abstract: A method includes storing a function representing a set of data elements stored in a backing memory and, in response to a first memory read request for a first data element of the set of data elements, calculating a function result representing the first data element based on the function.

Abstract: An apparatus and method for efficiently transmitting data are described. A transmitter sends data to a receiver. An encoder of the transmitter divides a received first block of data into multiple sub-blocks. The encoder selects a portion of each sub-block to compare to one another. A portion in a particular sub-block has a same offset and a same size as other portions of other sub-blocks. If the encoder determines the multiple portions match one another, the encoder sends, to the receiver, a second block of data corresponding to the first block of data. The second block of data has a same size as a size of the received first block of data, and the second block of data includes security data from one of multiple error correction schemes. Therefore, the second block of data provides security without increasing an amount of data to transmit.

Abstract: An electronic device includes a memory having a plurality of memory rows and a memory refresh functional block that performs a victim row refresh operation. For the victim row refresh operation, the memory refresh functional block selects one or more victim memory rows that may be victims of data corruption caused by repeated memory accesses in a specified group of memory rows near each of the one or more victim memory rows. The memory refresh functional block then individually refreshes each of the one or more victim memory rows.

Abstract: A processing system performs error detection at each of a plurality of layers of a neural network, such as a neural network implemented at a computational analog memory. By performing error detection at the layer level, the processing system is able to account for write errors when updating neural network weights, without waiting for backpropagation based on an output of the neural network. The processing system thereby reduces the amount of time needed to train the network, both by reducing the number of training epochs, and by reducing the length of the individual training epochs.

Abstract: A synergistic approach to mitigating crosstalk in a Dynamic Random-Access Memory (DRAM) implements the use of a random number generator to increment a counter in a probabilistic manner. The counter is formed by reclaiming bytes of a double data rate (DDR) fault isolation feature. The random number generator value may be compared against a predetermined parameter value and a determination may be made whether or not to extract and increment the counter based on a result of the comparison. A logic controller compares the counter value to a predetermined hotness threshold parameter and a flag is set based on an existence of an address match in local memory. Based on the results of the comparison, access to the DRAM is reduced.

Abstract: Systems, apparatuses, and methods for implementing a multi-tiered approach to cache compression are disclosed. A cache includes a cache controller, light compressor, and heavy compressor. The decision on which compressor to use for compressing cache lines is made based on certain resource availability such as cache capacity or memory bandwidth. This allows the cache to opportunistically use complex algorithms for compression while limiting the adverse effects of high decompression latency on system performance. To address the above issue, the proposed design takes advantage of the heavy compressors for effectively reducing memory bandwidth in high bandwidth memory (HBM) interfaces as long as they do not sacrifice system performance. Accordingly, the cache combines light and heavy compressors with a decision-making unit to achieve reduced off-chip memory traffic without sacrificing system performance.

Abstract: Systems, apparatuses, and methods for implementing a multi-tiered approach to cache compression are disclosed. A cache includes a cache controller, light compressor, and heavy compressor. The decision on which compressor to use for compressing cache lines is made based on certain resource availability such as cache capacity or memory bandwidth. This allows the cache to opportunistically use complex algorithms for compression while limiting the adverse effects of high decompression latency on system performance. To address the above issue, the proposed design takes advantage of the heavy compressors for effectively reducing memory bandwidth in high bandwidth memory (HBM) interfaces as long as they do not sacrifice system performance. Accordingly, the cache combines light and heavy compressors with a decision-making unit to achieve reduced off-chip memory traffic without sacrificing system performance.

Abstract: An electronic device includes a memory having a plurality of memory rows and a memory refresh functional block that performs a victim row refresh operation. For the victim row refresh operation, the memory refresh functional block selects one or more victim memory rows that may be victims of data corruption caused by repeated memory accesses in a specified group of memory rows near each of the one or more victim memory rows. The memory refresh functional block then individually refreshes each of the one or more victim memory rows.

Abstract: Data transmission mitigating interwire crosstalk including: dividing a data block to be transmitted from a transmitter to a receiver across a set of signal wires into sub-blocks; encoding each of the sub-blocks into a plurality of codewords; selecting, for each sub-block by a cost function, one of the codewords that is less likely to introduce interwire crosstalk; transmitting the selected codewords; and updating the cost function at the transmitter with feedback from the receiver.

Abstract: A method includes adding a set of one or more victim rows to a first probabilistic filter and to a second probabilistic filter, in response to a memory access request, identifying a candidate victim row adjacent to a memory address specified by a memory access request, identifying the candidate victim row as a victim row in the set of victim rows based on performing a lookup of the candidate victim row in a selected filter, where the selected filter includes one of the first probabilistic filter and the second probabilistic filter, in response to identifying the candidate row as the victim row, enabling a row hammering countermeasure, clearing the first probabilistic filter in each of a first set of time periods, and clearing the second probabilistic filter in each of a second set of time periods interleaved with the first set of time periods.

Abstract: Data transmission mitigating interwire crosstalk including: dividing a data block to be transmitted from a transmitter to a receiver across a set of signal wires into sub-blocks; encoding each of the sub-blocks into a plurality of codewords; selecting, for each sub-block by a cost function, one of the codewords that is less likely to introduce interwire crosstalk; transmitting the selected codewords; and updating the cost function at the transmitter with feedback from the receiver.

Abstract: A method includes receiving a write request for writing incoming data to a target memory line and, in response to the write request, comparing the incoming data with existing data in the target memory line to determine a number of a first type of state transition. The method further includes, in response to determining that the number of the first type of state transition for the write request exceeds a threshold, prior to writing the incoming data to the target memory line, storing adjacent data from each of a set of memory lines adjacent to the target memory line, and after writing the incoming data to the target memory line, writing the stored data to the set of adjacent memory lines.

Abstract: A method and apparatus of protecting a memory from a write attack includes dividing a cacheline of memory into a plurality of sub-blocks. A codeword is generated from at least one sub-block of the plurality of sub-blocks and a complement of the at least one sub-block. One of the generated codewords is selected, wherein the selected codeword is used for storage in memory.