Abstract: A method for operating a memory having a plurality of banks accessible in parallel, each bank including a plurality of grains accessible in parallel is provided. The method includes: based on a memory access request that specifies a memory address, identifying a set that stores data for the memory access request, wherein the set is spread across multiple grains of the plurality of grains; and performing operations to satisfy the memory access request, using entries of the set stored across the multiple grains of the plurality of grains.

Abstract: An integrated circuit includes a first terminal for receiving a data signal, a second terminal for receiving an external reference voltage, a receiver, and a reference voltage generation circuit. The receiver is powered by a power supply voltage with respect to ground and has a first input coupled to the first terminal, a second input for receiving a shared reference voltage, and an output for providing a data input signal. The reference voltage generation circuit is coupled to the second terminal and receives the power supply voltage. The reference voltage generation circuit is operable to form the shared reference voltage by mixing noise from the power supply voltage and noise from the second terminal.

Abstract: Systems and methods for cache replacement are disclosed. Techniques are described that determine a re-reference interval prediction (RRIP) value of respective data blocks in a cache, where an RRIP value represents a likelihood that a respective data block will be re-used within a time interval. Upon an access, by a processor, to a data segment in a memory, if the data segment is not stored in the cache, a data block in the cache to be replaced by the data segment is selected, utilizing a binary tree that tracks recency of data blocks in the cache.

Abstract: Methods, devices, and systems for information communication. Information transmitted from a host to a graphics processing unit (GPU) is received by information analysis circuitry of a field-programmable gate array (FPGA). A pattern in the information is determined by the information analysis circuitry. A predicted information pattern is determined, by the information analysis circuitry, based on the information. An indication of the predicted information pattern is transmitted to the host. Responsive to a signal from the host based on the predicted information pattern, the FPGA is reprogrammed to implement decompression circuitry based on the predicted information pattern. In some implementations, the information includes a plurality of packets. In some implementations, the predicted information pattern includes a pattern in a plurality of packets. In some implementations, the predicted information pattern includes a zero data pattern.

Abstract: Systems, apparatuses, and methods for implementing a multi-kernel wavefront scheduler are disclosed. A system includes at least a parallel processor coupled to one or more memories, wherein the parallel processor includes a command processor and a plurality of compute units. The command processor launches multiple kernels for execution on the compute units. Each compute unit includes a multi-level scheduler for scheduling wavefronts from multiple kernels for execution on its execution units. A first level scheduler creates scheduling groups by grouping together wavefronts based on the priority of their kernels. Accordingly, wavefronts from kernels with the same priority are grouped together in the same scheduling group by the first level scheduler. Next, the first level scheduler selects, from a plurality of scheduling groups, the highest priority scheduling group for execution. Then, a second level scheduler schedules wavefronts for execution from the scheduling group selected by the first level scheduler.

Abstract: A system and method for efficiently routing scan data between two dies used in three-dimensional packaging are described. In various implementations, a computing system includes at least a first semiconductor die (or first die) and a second die connected to one another within a three-dimensional (3D) package. The first die and the second die have multiple non-scan input/output (I/O) data channels between them for data transfer. The non-scan I/O data channels are partitioned into groups. The first die receives a given scan input data bit for testing a device under test (DUT) on the second die. The first die selects a first group of non-scan I/O data channels, and sends, to the second die, a copy of the given scan input data bit on each non-scan I/O data channel of the first group. The second die uses a voter circuit to determine the value of the given scan input data bit.

Abstract: A computing system may implement a basic input/output system (BIOS) update method. The BIOS also includes identifying an installed central processing unit (CPU) of a computer system coupled to a BIOS chipset, selecting CPU firmware corresponding to the installed CPU from a plurality of CPU platform firmware stored on a first memory, and loading the CPU firmware into a shared portion of a second memory coupled to the BIOS chipset, where the shared portion of the second memory is configured to store the CPU firmware as secondary CPU firmware.

Abstract: A device includes a communication scheduler to generate schedule trees for scheduling data communication among a plurality of nodes configured to perform a collective operation using data contributed from the plurality of nodes. The device includes data reduction logic to: identify one or more skewed nodes among the plurality of nodes, perform, according to a first set of schedule trees, a first operation to generate partial results based on data contributed from non-skewed nodes, and perform, according to a second set of schedule trees, a second operation to generate final results based on the partial results and data contributed from the one or more skewed nodes.

Abstract: A processing system that includes a shared data fabric resets a first client processor while operating a second client processor. The first client processor is instructed to stop making requests to one or more devices of the shared data fabric. Status communications are blocked between the first client processor and a memory controller, the second client processor, or both, such that the first client processor enters a temporary offline state. The first client processor is indicated as being non-coherent. Accordingly, when the processor is reset some errors and efficiency losses due messages sent during or prior to the reset are prevented.

Abstract: A technique for operating a cache is disclosed. The technique includes based on a workload change, identifying a first allocation permissions policy; operating the cache according to the first allocation permissions policy; based on set sampling, identifying a second allocation permissions policy; and operating the cache according to the second allocation permissions policy.

Abstract: A method for receiving a multi-level error signal having more than two logic levels includes oversampling the multi-level error signal to provide sampled symbols, wherein a first level of the multi-level error signal indicates no error, and second and third levels of the multi-level error signal indicate first and second error conditions, respectively. The sampled signals are de-serialized to provide sets of symbols. A start of a symbol period is determined in response to detecting that a given sample is different from a prior sample, and the prior sample indicates no error. The sets of symbols are filtered to provide corresponding output symbols based on the start.

Abstract: Performance adaptation for an integrated circuit includes receiving, by a workload prediction system of a hardware processor, telemetry data for one or more systems of the hardware processor. A workload prediction is determined by processing the telemetry data through a workload prediction model executed by a workload prediction controller of the workload prediction system. A profile is selected, from a plurality of profiles, that matches the workload prediction. The selected profile specifies one or more operating parameters for the hardware processor. The selected profile is provided to a power management controller of the hardware processor for controlling an operational characteristic of the one or more systems.

Abstract: A framework disclosed herein extends a relaxed, scoped memory model to a system that includes nodes across a commodity network and maintains coherency across the system. A new scope, cluster scope, is defined, that allows for memory accesses at scopes less than cluster scope to operate on locally cached versions of remote data from across the commodity network without having to issue expensive network operations. Cluster scope operations generate network commands that are used to synchronize memory across the commodity network.

Abstract: A processing system includes a first processor couplable to a first memory and a second memory. In response to a page migration trigger for a page in the first memory, the first processor is configured to, responsive to the page being a read-only page storing code for execution, initiate migration of the page to a code cache portion of a second memory associated with a second processor and shared by multiple processes executing at the second processor, and to configure each process of a set of processes executing at the second processor to access and execute the code from the code cache portion.

Abstract: Methods and systems are disclosed for transitioning, by a hardware-based controller, a system on a chip (SoC) into different power states. Techniques disclosed include tracking, by the controller, metrics associated with the SoC and transitioning, by the controller, the SoC from a first power state to a second power state based on the tracked metrics. Were the total amount of power that is used by at least a portion of the transition between the first power state to the second power state and a time spent in the second power state is less than the total amount of power that would have been used by remaining in the first power state.

Abstract: Systems and methods related to run-time selection of a render mode in which to execute command buffers with a graphics processing unit (GPU) of a device based on performance data corresponding to the device are provided. A user mode driver (UMD) or kernel mode driver (KMD) executed at a central processing unit (CPU) selects abinning mode based on whether performance data that includes sensor data or performance counter data indicates that an associated binning condition or override condition has been met. The UMD or the KMD causes pending command buffers to be patched to execute in the selected binning mode based on whether the binning mode is enabled or disabled.

Abstract: A memory sprint controller, responsive to an indicator of an irregular memory access phase, causes a memory controller to enter a sprint mode in which it temporarily adjusts at least one timing parameter of a dynamic random access memory (DRAM) to reduce a time in which a designated number of activate (ACT) commands are allowed to be dispatched to the DRAM.

Abstract: An input/output memory management unit includes a control logic circuit and a device table entry valid bit array. The control logic circuit provides physical addresses in response to virtual addresses of memory access requests from a plurality of input/output devices. The device table entry valid bit array stores a plurality of valid bits corresponding to different ones of the plurality of input/output devices. The control logic circuit accesses a first valid bit corresponding to a first input/output device from the device table entry valid bit array, and selectively accesses a device table in a system memory in response to a state of the valid bit.

Abstract: Methods and systems are disclosed for clock delay compensation in a multiple chiplet system. Techniques disclosed include distributing, by a clock generator, a clock signal across distribution trees of respective chiplets; measuring phases, by phase detectors, where each phase measurement is associated with a chiplet of the chiplets and is indicative of a propagation speed of the clock signal through the distribution tree of the chiplet. Then, for each chiplet, techniques are further disclosed that determine, by a microcontroller, based on the phase measurements associated with the chiplet, a delay offset, and that delay, based on the delay offset, the propagation of the clock signal through the distribution tree of the chiplet using a delay unit associated with the chiplet.

Abstract: A processing unit, method, and medium for decompressing or generating textures within a graphics processing unit (GPU). The textures are compressed with a variable-rate compression scheme such as JPEG. The compressed textures are retrieved from system memory and transferred to local cache memory on the GPU without first being decompressed. A table is utilized by the cache to locate individual blocks within the compressed texture. A decompressing shader processor receives compressed blocks and then performs on-the-fly decompression of the blocks. The decompressed blocks are then processed as usual by a texture consuming shader processor of the GPU.