Abstract: Disclosed herein are methods, apparatuses, and systems for accessing vertex data stored in a memory, and applications thereof. Such a method includes writing vertex data of primitives into contiguous banks of a memory such that the vertex data of consecutively written primitives spans more than one row of the memory. Vertex data of two consecutively written primitives are read from the memory in a single clock cycle.

Abstract: A prefetch device and method are disclosed that determines from which addresses to speculatively fetch data based on information collected regarding previous cache-miss addresses. A historical record showing a propensity to experience cache-misses at a particular address-offset from a prior cache-miss address within a region of memory provides an indication that data needed by future instructions has an increased likelihood to be located at a similar offset from a current cache-miss address. The prefetch device disclosed herein maintains a record of the relationship between a cache-miss address and subsequent cache-miss addresses for the most recent sixty-four unique data manipulation instructions that resulted in a cache-miss. The record includes a weighted confidence value indicative of how many cache-misses previously occurred at each of a selection of offsets from a particular cache-miss address.

Abstract: Methods and apparatuses are provided for controlling power consumption in a processor (or computational unit thereof). The method comprises monitoring power consumption in a processor (or computational unit) and determining that the power consumption of the processor (or computational unit) exceeds a threshold. Thereafter, instruction issuance if modified (such as by slowing or ceasing instruction issuance) within the processor (or computational unit) until the power consumption is below the threshold. The apparatus comprises a power consumption monitor for determining when power consumption within the processor exceeds a threshold. Upon that determination, a scheduler begins modify instruction issuance to one or more execution units until the power consumption is below the threshold. The modification of instruction issuance can be to slow instruction issuance or cease instruction issuance for a time period or until the power consumption is below the threshold.

Abstract: A method, system, and computer program product are provided for adjusting write timing in a memory device based on a training signal. For instance, the method can include configuring the memory device in a training mode of operation. The method can also include determining a write timing window between a signal on a data bus and a write clock signal based on the training signal. Further, the method includes adjusting a phase difference between the signal on the data bus and the write clock signal based on the write timing window. The memory device can recover data on the data bus based on the adjusted phase difference.

Abstract: A method and apparatus are provided for increasing the accuracy of a branch predictor. A branch prediction table provides a first instance of a branch prediction value associated with an instruction being speculatively executed a first time; and provides a second instance of the branch prediction value associated with the instruction being speculatively executed a second rime. The first instance of the branch prediction value may be subsequently revised after the instruction associated with the first instance of the branch prediction value is retired. Information regarding whether that branch instruction was accurately predicted may then be used to update the branch prediction table and the second instance of the branch prediction value.

Abstract: Techniques are disclosed relating to detecting and minimizing timing problems created by clock domain crossing (CDC) in integrated circuits. In various embodiments, one or more timing parameters are associated with a path that crosses between clock domains in an integrated circuit, where the one or more timing parameters specify a propagation delay for the path. In one embodiment, the timing parameters may be distributed to different design stages using a configuration file. In some embodiments, the one or more parameters may be used in conjunction with an RTL model to simulate propagation of a data signal along the path. In some embodiments, the one or more parameters may be used in conjunction with a netlist to create a physical design for the integrated circuit, where the physical design includes a representation of the path that has the specified propagation delay.

Abstract: A memory cell system is provided forming a first insulator layer over a semiconductor substrate, forming a charge trap layer over the first insulator layer, forming an intermediate layer over the charge trap layer, and forming a second insulator layer with the intermediate layer.

Abstract: A resonant clock network includes an inductor coupled to the clock network through a plurality of switches. When the clock network enters resonant mode, the turn-on of the switches to couple the inductor to the clock network is staggered. The clock network may be formed of multiple regions, each with its own inductor and switches. The turn-on of switches of each region may be staggered with respect to the turn-on off the switches of the other regions as well as to the turn-on of switches within a region. In addition to staggering the turn-on of the switches when entering the resonant mode, the switches may be turned off in a staggered manner when exiting the resonant mode of operation.

Abstract: A system and method for efficient data prefetching. A data stream stored in lower-level memory comprises a contiguous block of data used in a computer program. A prefetch unit in a processor detects a data stream by identifying a sequence of storage accesses referencing a contiguous blocks of data in a monotonically increasing or decreasing manner. After a predetermined training period for a given data stream, the prefetch unit prefetches a portion of the given data stream from memory without write permission, in response to an access that does not request write permission. Also, after the training period, the prefetch unit prefetches a portion of the given data stream from lower-level memory with write permission, in response to determining there has been a prior access to the given data stream that requests write permission subsequent to a number of cache misses reaching a predetermined threshold.

Abstract: Techniques are disclosed relating to improving the performance of branch prediction in processors. In one embodiment, a processor is disclosed that includes a branch prediction unit configured to predict a sequence of instructions to be issued by the processor for execution. The processor also includes a pattern detection unit configured to detect a pattern in the predicted sequence of instructions, where the pattern includes a plurality of predicted instructions. In response to the pattern detection unit detecting the pattern, the processor is configured to switch from issuing instructions predicted by the branch prediction unit to issuing the plurality of instructions. In some embodiments, the processor includes a replay unit that is configured to replay fetch addresses to an instruction fetch unit to cause the plurality of predicted instructions to be issued.

Abstract: A method, computer program storage device and system are provided for determination and selection of optimized circuit components. The method includes performing a timing analysis on at least a portion of an electronic circuit and determining a path in the at least a portion of an electronic circuit, where the path comprises at least one storage element and an operational attribute associated with the path. The method also includes determining an optimized storage element adapted to utilize the operational attribute. The system includes a processing device and at least one of a synthesis tool, a timing tool or a place and route tool communicatively connected to the processing device. The synthesis tool, the timing tool and the place and route tool are adapted to process or analyze an electrical circuit.

Abstract: A multi-core microprocessor provides an indication of the power management state of each of the cores on output terminals. Cooling of the cores is adjusted responsive to the indication of the power management state of the respective cores with additional cooling being provided to those cores in a more active state and less cooling provided to those cores in a less active state.

Abstract: A system, method and article of manufacture for an accelerated processing device (APD) to request a central processing unit (CPU) to process a task, comprising enqueuing a plurality of tasks on a queue using the APD, generating a user-level interrupt and transmitting to the CPU the plurality of tasks in the queue using an interrupt handler associated with a CPU thread.

Abstract: Some embodiments of a processing system implement a memory-controller-parallelism-aware scheduling technique. In at least one embodiment of the invention, a method of operating a processing system includes scheduling a memory request requested by a thread of a plurality of threads executing on at least one processor according to thread priority information associated with the plurality of threads. The thread priority information is based on a maximum of a plurality of local memory bandwidth usage indicators for each thread of the plurality of threads. Each of the plurality of local memory bandwidth usage indicators for each thread corresponds to a respective memory controller of a plurality of memory controllers.

Abstract: In a processing system comprising a plurality of processing nodes coupled via a switching fabric, a method includes implementing a flow control property for a data flow in the switching fabric based on an addressing property of an address of a virtual network interface controller associated with the data flow. A switching fabric includes a plurality of ports, each port coupleable to a corresponding processing node, and switching logic coupled to the plurality of ports. The switching fabric further includes flow control logic to implement a flow control property for a data flow in the switching logic based on an addressing property of an address of a virtual network interface controller associated with the data flow.

Abstract: A system and method for detecting errors in high-speed asymmetric interfaces are described. Embodiments include transmitting digital data between a first system component and a second system component over a bidirectional interface, wherein the first component is significantly more intelligent than the second component. The first component controls many operations of the second component, including receiving a signature from the second component over an existing line of the interface. The signature received is compared to a signature stored by the first component. Both signatures correspond to a transaction over the interface. Based on the comparison, the first component determines whether the transaction was successful, and directs the second component as necessary.

Abstract: A method and apparatus to improve the efficiency of debugging a processor is provided. Also provided is a computer readable storage device encoded with data for adapting a manufacturing facility to create an apparatus. The method includes receiving a first test data, which identifies a state of a state machine, wherein the state machine performs reset and initialization operations for a processor. The method also includes halting the state machine in the state identified by the first test data upon reaching the state.

Abstract: A method for switching decoding and rendering of a digital video stream from a first graphics processing unit (GPU) to a second GPU. The digital video stream is evaluated to determine an amount of time until a next intra-coded frame (I-frame) in the digital video stream. If the amount of time is below a threshold, decoding and rendering of the digital video stream is switched to the second GPU on the next I-frame in the digital video stream and decoding the digital video stream by the first GPU is stopped. If the amount of time is above the threshold, the digital video stream is decoded on both the first GPU and the second GPU, the rendering of the digital video stream is switched to the second GPU, and decoding the digital video stream by the first GPU is stopped.

Abstract: A method and apparatus are provided for controlling system management interrupts is disclosed. An interrupt filter comprises a memory, a comparator and a logic circuit. The memory is adapted to contain a list indicating one or more devices with permission associated with an interrupt signal. The comparator is adapted to receive an interrupt signal containing type information from the one or more devices. The comparator is adapted to compare the interrupt type against the list to determine if the one or more devices is permitted to send the interrupt signal. The logic circuit blocks or passes the interrupt signal in response to the result of the comparison.

Abstract: Systems and methods describe herein provide a method of for managing task scheduling on a accelerated processing device. Duration characteristics for a plurality of offset values are determined based on execution of first and second processing tasks within an accelerated processing device. An offset value from the plurality of offset values is selected indicating a difference in an execution start time between the first processing task and the second processing task. Additional executions of the first and second processing tasks are scheduled based on the selected offset value.