Abstract: With the success of programming models such as OpenCL and CUDA, heterogeneous computing platforms are becoming mainstream. However, these heterogeneous systems are low-level, not composable, and their behavior is often implementation defined even for standardized programming models. In contrast, the method and system embodiments for the heterogeneous parallel primitives (HPP) programming model disclosed herein provide a flexible and composable programming platform that guarantees behavior even in the case of developing high-performance code.

Abstract: A tunnel for a communication system includes first and second bridges. The first bridge has a first port adapted to couple to a first link and a second port, and has a first programmable bus number and a first programmable function number. The second bridge has a first port coupled to the second port of the first bridge, and a second port, and has a second programmable bus number and a second programmable function number. In a hoist enabled mode, the first bridge forwards a packet on the first link to the second bridge if the second programmable bus number is equal to the first programmable bus number, a bus number of the packet is equal to the first programmable bus number, and a function number of the packet is equal to the second programmable function number.

Abstract: A system and method for providing a docking station that supports bi-directional high speed data, high bandwidth display, and power to a computing device utilizing a standard connector on the computing device are described. This includes a standard connector on the computing device including a standard digital display connector having a first set of two lanes and a second set of two lanes, a USB host that provides USB signals that enable bi-directional high speed data, a digital display source that provides digital display signals that enable high bandwidth display and couples digital display signals to the digital display connector on the second set of lanes, a multiplexor that receives signals from the USB host, receives signals from the digital display source, and couples the USB signals to the digital display connector on the first set of lanes, and a power subsystem that receives power via the digital display connector.

Abstract: A memory loopback system and method including an address/command transmit source configured to transmit a command and associated address through an address/command path. A transmit data source is configured to transmit write data associated with the command through a write path. Test control logic is configured to generate gaps between successive commands. A loopback connection is configured to route the write data from the write path to a read path. A data comparator is configured to compare the data received via the read path to a receive data source and generate a data loopback status output. Pattern generation logic can be configured to generate a loopback strobe, the loopback strobe being coupled to the read path. The pattern generation logic may be configured to synthesize a read strobe based on the test control logic and to use the synthesized read strobe as the loopback strobe.

Abstract: An integrated circuit (IC) chip having repeaters for propagating signals along relatively long wires that extend between and among lower-level physical blocks of the IC chip, wherein the repeaters are implemented as clocked flip-flops (or “repeater flops”). A method for automatically inserting and allocating such repeater flops during the logical and physical design of the IC chip is also provided.

Abstract: A method is provided for controlling a data transmission device. The method includes providing a reference voltage to the common mode driver and putting the data transmission device in a low power state. The method also includes driving a differential signal pair output from the common mode driver during a portion of the low power state. Also provided is a device that includes a data output driver portion configured to drive an output signal at a common mode voltage and a data output driver portion configured to drive an output signal at a differential voltage level during at least a portion of time when the device is not in a low power state. Also provided is a computer readable storage device encoded with data for adapting a manufacturing facility to create the device. Also provided is an apparatus configured to perform the method.

Abstract: A cache is configured to have a first cache line allocation policy for a memory address. An instruction associated with the memory address is received and a second cache line allocation policy is determined based on the instruction. The cache is reconfigured to have the second cache line allocation policy in response to receiving the instruction. A data processor includes processor core to receive and execute an instruction associated with a memory address, a cache including a plurality of cache lines, and a cache allocation module to determine a cache line allocation policy based on the instruction and to reconfigure the cache to have the cache line allocation policy for execution of the instruction at the processor core.

Abstract: A system and method for calibrating bias in a data transmission system including a calibrated bias having impedance calibration for accommodating parameter variations in the data transmission system. A current mirror receives and balances bias currents between the calibrated bias and an output driver from the data transmission system. A digital compensation logic circuit is connected to the calibrated bias to adjust the calibrated bias for variations in parameters causing a current tail effect. A calibration logic circuit adjusts calibration due to variations in operational parameters, such that the tail current variations are minimized.

Abstract: A method and apparatus for partitioning a cache includes determining an allocation of a subcache out of a plurality of subcaches within the cache for association with a compute unit out of a plurality of compute units. Data is processed by the compute unit, and the compute unit evicts a line. The evicted line is written to the subcache associated with the compute unit.

Abstract: Method, system, and computer program product embodiments for synchronizing workitems on one or more processors are disclosed. The embodiments include executing a barrier skip instruction by a first workitem from the group, and responsive to the executed barrier skip instruction, reconfiguring a barrier to synchronize other workitems from the group in a plurality of points in a sequence without requiring the first workitem to reach the barrier in any of the plurality of points.

Abstract: Disclosed methods, systems, and computer program products embodiments include synchronizing a group of workitems on a processor by storing a respective program counter associated with each of the workitems, selecting at least one first workitem from the group for execution, and executing the selected at least one first workitem on the processor. The selecting is based upon the respective stored program counter associated with the at least one first workitem.

Abstract: A decimal floating-point (DFP) adder includes a decimal leading-zero anticipator (LZA). The DFP adder receives DFP operands. Each operand includes a significand, an exponent, a sign bit and a leading zero count for the significand. The DFP adder adds or subtracts the DFP operands to obtain a DFP result. The LZA determines the leading zero count associated with the significand of the DFP result. The LZA operates at least partially in parallel with circuitry (in the DFP adder) that computes the DFP result. The LZA does not wait for that circuitry to finish computation of the DFP result. Instead it “anticipates” the number of leading zeros that the result's significand will contain.

Abstract: A method for forming a memory device is provided. A nitride layer is formed over a substrate. The nitride layer and the substrate are etched to form a trench. The nitride layer is trimmed on opposite sides of the trench to widen the trench within the nitride layer. The trench is filled with an oxide material. The nitride layer is stripped from the memory device, forming a mesa above the trench.

Abstract: A method of controlling voltage in a circuit is provided. Within the circuit, a block of an electrical component provides an indication that it desires to switch states (such as from off to on, on to off, or from one speed to another). The change in states requires a different current draw by the electrical component block. The indication is received by an electrical component that controls the voltage of the circuit. The electrical component that controls the voltage then issues a signal granting permission for the electrical component block to switch states. This permission signal is received by the electrical component and the electrical component block changes state.

Abstract: Information of a first type is determined at an integrated circuit die of a data processing device included an integrated circuit package. The integrated circuit package includes the first integrated circuit die and a second integrated circuit die. Information of a second type is determined at the integrated circuit die. The first and second type of information is transmitted from the integrated circuit die to another integrated circuit die using a time-divided multiplexed protocol by transmitting the first information during a first time slot of the protocol and transmitting the second information during a second time slot of the protocol.

Abstract: The present invention provides a system for handling extra contexts for shader constants, and applications thereof. In an embodiment there is provided a computer-based method for executing a series of compute packets in an execution pipeline. The execution pipeline includes a first plurality of registers configured to store state-updates of a first type and a second plurality of registers configured to store state-updates of a second type. A first number of state-updates of the first type and a second number of state-updates of the second type are respectively identified and stored in the first and second plurality of registers. A compute packet is sent to the execution pipeline responsive to the first number and the second number. Then, the compute packet is executed by the execution pipeline.

Abstract: An integrated circuit includes a clock-tree with a plurality of clock buffers, a plurality of clocked storage elements, and a plurality of logic circuits. Each clocked storage element has a clock input terminal connected to one of the plurality of clock buffers and a weight. Each of the logic circuits is associated with two of the plurality of clocked storage elements and is characterized as having a logic depth. The weight of each clocked storage element is equal to a sum of an inverse of a logic depth of each of the plurality of logic circuits associated therewith. A first clocked storage element which has a highest weight and is adjacent to and interacts with a second clocked storage element via one of the plurality of logic circuits. A first clock buffer provides a common clock signal to the first and second clocked storage elements.

Abstract: A peripheral device can be powered off when not in use by redirecting accesses to the peripheral device's configuration space from the peripheral device to a memory located separate from the peripheral device. A method for redirecting accesses includes copying the current contents of the configuration space to the memory. Accesses to the configuration space are redirected to the memory, whereby the memory services the accesses to the configuration space. After the redirection is enabled, the peripheral device can be powered off. When the peripheral device needs to be used again, it is powered on and the contents of the memory are copied to the configuration space. The configuration space can then resume servicing configuration space accesses.

Abstract: Described are a circuit and a method of analyzing and correcting a fault occurring in operation of the circuit during a power gating sequence. The method includes executing a modification of the power gating sequence that includes maintaining operation of a trace capture buffer (TCB); recording, in the TCB, events occurring during the executing; and correcting the fault based on analysis of the events recorded in the TCB. The circuit includes a plurality of components including a TCB, and a switch configured to maintain power to the TCB in a first state and turn off power to the TCB in a second state.

Abstract: Power supply variations and jitter are measured by monitoring the performance of a ring oscillator on a cycle-by-cycle basis. Performance is measured by counting the number of stages of the ring oscillator that are traversed during the clock cycle and mapping the number of stages traversed to a particular voltage level. Counters are used to count the number of ring oscillator revolutions and latches are used to latch the state of the ring oscillator at the end of the cycle. Based on the counters and latches, a monitor output is generated that may also incorporate an adjustment for a reset delay associated with initializing the ring oscillator and counters to a known state.