Abstract: An apparatus for detecting the presence of assembly related defects on a semiconductor device including an edge ring having a resistance value and including one or more layers configured to at least partially cover the semiconductor device in a first direction. The one or more layers are divided into a first section and a second section. Each layer of the one or more layers are in electrical communication with one another. The resistance value of the edge ring is at a first resistance value associated with the first and second sections being intact. At least one of the first section and second section is configured to break in response to an assembly related defect, and the resistance value of the edge ring is configured to change from the first resistance value to a second resistance value in response to at least one of the first and second sections being broken.

Abstract: A semiconductor memory package includes a substrate, a first stack of memory dies, and a second stack of memory dies. The substrate includes a top layer and a bottom layer. The first stack of memory dies is electrically coupled to the top layer of the substrate and includes a controller and a first number of memory dies. The second stack of memory dies is electrically coupled to the top layer of the substrate and includes a second number of memory dies greater than the first number of memory dies. An upper surface of the first stack of memory dies and an upper surface of the second stack of memory dies are substantially coplanar.

Abstract: Multi-signal vias for use with differential pair signals in electronic devices. The electronic devices include a printed circuit board having a first side and a second side opposite the first side, a first conductive trace on the first side of the substrate and a second conductive trace on the first side of the substrate. The printed circuit board also includes a shared via, which includes a first conductive portion and a second conductive portion. The first conductive portion and the second conductive portion are separated by a non-conductive portion. The first conducive trace is coupled to the first conductive portion of the shared via and the second conductive trace is coupled to the second conductive portion of the shared via.

Abstract: Disclosed herein is a data storage device. A data port transmits data between a host computer system and the data storage device. A non-volatile storage medium stores encrypted user content data and a cryptography engine uses a cryptographic key to decrypt the encrypted user content data stored on the storage medium in response to a request from the host computer system. An access controller receives a request from a manager device to initialize the data storage device. The controller generates the cryptographic key, generates a manager key configured to provide manager access for the manager device and provide access to the cryptographic key, and stores, on a data store, authorization data indicative of the manager key and accessible based on a private key stored on the manager device.

Abstract: A testing apparatus for Devices Under Test (DUTs) includes at least one intake damper and at least one exhaust damper. At least one fan moves recirculated fluid and exterior fluid across one or more DUTs inside the testing apparatus. In one aspect, the testing apparatus includes a door to provide access to a chamber and the door includes at least one channel. At least a portion of the fluid flows through the at least one channel of the door. In another aspect, the door is configured to provide access to a chamber from the front of the chamber and the fluid is moved in a direction across the one or more DUTs substantially from the front of the chamber towards a rear of the chamber.

Abstract: Methods and apparatus are provided for arbitrating access to, and usage of, various device resources of a data storage device (DSD) configured for Machine Learning with Low-Power. The data storage device may include a TinyML controller with an artificial intelligence (AI) accelerator integrated with a data storage controller on a system-on-a-chip (SoC). The device resources may be, e.g., storage resources such as random access memory (RAM) devices, non-volatile memory (NVM) arrays, and latches formed on NVM dies of the NVM arrays. The resource arbitration may be based, for example, on parameters pertaining to ML operations performed by an ML controller that includes the AI accelerator, such as a turnaround time of an ML epoch or a stage-wise execution time. The resource arbitration is configured to provide for the efficient interleaving of the ML/AI operations performed by the ML controller and data storage operations performed by the data storage controller.

Abstract: Aspects of a storage device are provided which apply advanced thermal throttling in response to temperature changes based on multiple thermal power states for different types of cells, such as SLCs and MLCs. Initially, a controller determines that a temperature of the memory meets a thermal throttling threshold of a plurality of thermal throttling thresholds. Subsequently, the controller transitions into a thermal power state of a plurality of thermal power states when the temperature meets the thermal throttling threshold. The controller applies a thermal mitigation configuration associated with the thermal power state. The controller then determines that the temperature of the memory has reached a thermal equilibrium in the thermal power state based on the thermal mitigation configuration. Storage device performance is thus improved through advanced thermal throttling without compromising data integrity.

Abstract: A storage system has a memory with a multi-level cell (MLC) block that can store multiple bits per cell or can be constrained to store only one bit per cell. Using the MLC block to store only one bit per cell can increase the performance of the storage system but can also reduce endurance of the MLC block. The storage system can monitor a command queue to determine the performance needed. With that information, the storage system can determine whether it is worth making the tradeoff of increasing performance at the cost of endurance.

Abstract: A thermal surrogate device is configured to replace a data storage device within a server or other multiple data storage device system. The thermal surrogate device has a housing that has a form factor identical to the data storage device. The housing has a cavity along a length, and a sensor fixture is positioned within the housing. An airflow sensor is attached to the sensor fixture and configured to measure an airflow through the housing. The housing is configured to be installed within a slot for a data storage device in a multiple memory system.

Abstract: A data storage device includes a memory device and a controller coupled to the memory device. The controller is configured to determine that a hint calibration operation is needed, select a first hint mode out of a plurality of hint modes, generate one or more hints based on a selected hint mode, and select a hint mode based on one or more of a performance, quality of service, and power consumption of the data storage device. The controller is further configured to iterate through the plurality of hint modes during the hint calibration operation and operate based on the selected hint mode until the controller determines that another hint calibration operation is needed.

Abstract: A storage device is provided that performs constant biasing in priority blocks, such as OTP memory blocks (fuse ROM) and flash memory blocks having a threshold number of P/E cycles. The storage device includes an OTP memory, a flash memory, and a controller. The OTP memory includes a block having a word line and a plurality of cells coupled to the word line. The flash memory includes another block having a word line and a plurality of cells coupled to this word line. The controller is configured to apply a constant bias to the word line of the OTP memory block and, in some cases to the word line of the flash memory block, between execution of host commands. As a result, lower bit error rates due to wider Vt margins may occur while system power may be saved through selective application of constant biasing.

Abstract: Various illustrative aspects are directed to a data storage device, comprising: one or more disks; an actuating mechanism comprising one or more heads, and configured to position the one or more heads proximate to disk surfaces of the one or more disks; and one or more processing devices. The one or more processing devices are configured to: determine a first burst value based on an averaged value of a first set of one or more bursts; determine a second burst value based on an averaged value of a second set of one or more bursts; generate a position error signal (PES) based on the determined first burst value and the determined second burst value; and control a position of at least one head among the one or more heads based on the PES.

Abstract: The present disclosure generally relates to an efficient manner of fetching data for write commands. The data can be fetched prior to classification, which is a fetch before mode. The data can alternatively be fetched after classification, which is a fetch after mode. When the data is fetched after classification, the write commands are aggregated until sufficient data associated with any command is split between memory devices. When in fetch before mode, the data should properly align such that data associated with any command is not split between memory devices. Efficiently toggling between the fetch before and fetch after modes will shape how writes are performed without impacting latency and bandwidth without significantly increasing write buffer memory size.

Abstract: The present disclosure generally relates to estimating when data to be written will be read or re-written prior to actually writing the data to the memory device. The estimating can be used to smartly route the data to the appropriate memory location at the writing stage or to evict the data from a hot memory location to a colder memory location. To perform the estimating, typical traces or data may be used as may the metadata of the data. Separating data according to the data “temperature” (i.e. the expected access time and frequency), and usage to optimize the SLC partition usage has meaningful impact on several storage metrics such as performance and endurance.

Abstract: The present disclosure generally relates to a flip chip assembly having a bump that reduces stress levels in a low-k dielectric layer in the flip chip. Rather than having a single, large area plateau that interfaces with a large corresponding opening of an insulating layer in the flip chip, the bump includes a plurality of much smaller pillars that interface with a corresponding plurality of openings in the insulating layer. In so doing, the low-k layer within the flip chip experiences much less stress and hence, fewer failures.

Abstract: A storage device includes multiple memory dies and a controller configured to: (i) perform XOR parity computations for parity bins based, at least in part, on updated contents of a first user data memory cell and contents of each user data memory cell also assigned to the first parity bin, (ii) storing the first parity data into a first parity memory cell associated with the first parity bin; (iii) identify a second parity memory cell for dynamic reconfiguration based, at least in part, on performance data of the non-volatile memory device, the second parity memory cell being assigned to a second parity bin; (iv) copy the second parity memory cell to a third memory cell of the plurality of memory cells; and (v) associate the third memory cell with the second parity bin, thereby making the third memory cell a parity memory cell of the plurality of parity memory cells.

Abstract: The present disclosure generally relates to detecting command identification (CID) collisions in host commands. Host commands stored in submission queues are supposed to have unique CIDs. The host device selects the CID and attaches the CID to the command. Once the command is executed, the host device may reuse the CID. Sometimes, the host device reuses a CID before a command already using the CID is executed, which is a collision. Rather than search all CIDs to find a collision, redundancy bits can be created for each command, and the redundancy can be the same for multiple pending commands. The redundancy bits can be checked first to see if there is a match, followed by comparing CIDs for only those commands that have matching redundancy bits. In so doing, CID collisions are detected earlier and easier.

Abstract: A storage device is provided that allows a controller to directly access bytes of data in data latches connected to memory, as opposed to through controller RAM. The storage device may include a memory, a plurality of data latches connected to the memory, and a controller coupled to each of the data latches. The controller is configured to access one or more bytes of decoded data in one or more of the data latches. For instance, the controller may provide a command including an address for data in the memory, and may process one or more bytes of the data in at least one of the data latches in response to the command. The controller may also store a mapping of addresses for each of the word lines, including the address provided in the command. As a result, operation latency may be reduced and controller RAM savings achieved.

Abstract: Various illustrative aspects are directed to a data storage device comprising a first spindle motor configured to rotate one or more disks in a first stack of disks, a second spindle motor configured to rotate one or more disks in a second stack of disks, and one or more processing devices configured to detect back electromotive force (BEMF) voltages generated by the first spindle motor and the second spindle motor. In other aspects the one or more processing devices can control speeds of the first spindle motor and the second spindle motor based on the detected BEMF voltages.

Abstract: A storage system has a memory with memory cells that can store a non-power-of-two number of states. A map is used to distribute data bits in the memory. The map can be a modified version of a quadrature amplitude modulation (QAM) map. The mapping can be done by a controller in the storage system or by the memory die. Performing the mapping in the memory die can reduce data traffic between the controller and the memory die, which can provide an improvement to performance and power consumption.