Abstract: Unavoidable physical phenomena, such as an alpha particle strikes, can cause soft errors in integrated circuits. Materials that emit alpha particles are ubiquitous, and higher energy cosmic particles penetrate the atmosphere and also cause soft errors. Some soft errors have no consequence, but others can cause an integrated circuit to malfunction. In some applications (e.g. driverless cars), proper operation of integrated circuits is critical to human life and safety. To minimize or eliminate the likelihood of a soft error becoming a serious malfunction, detailed assessment of individual potential soft errors and subsequent processor behavior is necessary. Embodiments of the present disclosure facilitate emulating a plurality of different, specific soft errors. Resilience may be assessed over the plurality of soft errors and application code may be advantageously engineered to improve resilience.

Abstract: Graphics processing unit (GPU) performance and power efficiency is improved using machine learning to tune operating parameters based on performance monitor values and application information. Performance monitor values are processed using machine learning techniques to generate model parameters, which are used by a control unit within the GPU to provide real-time updates to the operating parameters. In one embodiment, a neural network processes the performance monitor values to generate operating parameters in real-time.

Abstract: Graphics processing unit (GPU) performance and power efficiency is improved using machine learning to tune operating parameters based on performance monitor values and application information. Performance monitor values are processed using machine learning techniques to generate model parameters, which are used by a control unit within the GPU to provide real-time updates to the operating parameters. In one embodiment, a neural network processes the performance monitor values to generate operating parameters in real-time.

Abstract: Unavoidable physical phenomena, such as an alpha particle strikes, can cause soft errors in integrated circuits. Materials that emit alpha particles are ubiquitous, and higher energy cosmic particles penetrate the atmosphere and also cause soft errors. Some soft errors have no consequence, but others can cause an integrated circuit to malfunction. In some applications (e.g. driverless cars), proper operation of integrated circuits is critical to human life and safety. To minimize or eliminate the likelihood of a soft error becoming a serious malfunction, detailed assessment of individual potential soft errors and subsequent processor behavior is necessary. Embodiments of the present disclosure facilitate emulating a plurality of different, specific soft errors. Resilience may be assessed over the plurality of soft errors and application code may be advantageously engineered to improve resilience.

Abstract: Unavoidable physical phenomena, such as an alpha particle strikes, can cause soft errors in integrated circuits. Materials that emit alpha particles are ubiquitous, and higher energy cosmic particles penetrate the atmosphere and also cause soft errors. Some soft errors have no consequence, but others can cause an integrated circuit to malfunction. In some applications (e.g. driverless cars), proper operation of integrated circuits is critical to human life and safety. To minimize or eliminate the likelihood of a soft error becoming a serious malfunction, detailed assessment of individual potential soft errors and subsequent processor behavior is necessary. Embodiments of the present disclosure facilitate emulating a plurality of different, specific soft errors. Resilience may be assessed over the plurality of soft errors and application code may be advantageously engineered to improve resilience.

Abstract: Integrated circuits, or computer chips, typically include multiple hardware components (e.g. memory, processors, etc.) operating under a shared power (e.g. thermal) constraint that is sourced by one or more power sources for the chip. Typically, the hardware components can be individually configured to operate at certain states (e.g. to operate at a certain frequency by setting a clock speed for a clock dedicated to the hardware component). Thus, each hardware component can be configured to operate at an operating point that is determined to be optimal, usually in terms of achieving some desired goal for a specific application (e.g. frame rates for gaming, etc.). In the context of chip hardware that operates under a shared power/thermal constraint, a method, computer readable medium, and system are provided for determining the optimal operating point for the chip that takes into consideration both performance of the chip and power consumption by the chip.

Abstract: Graphics processing unit (GPU) performance and power efficiency is improved using machine learning to tune operating parameters based on performance monitor values and application information. Performance monitor values are processed using machine learning techniques to generate model parameters, which are used by a control unit within the GPU to provide real-time updates to the operating parameters. In one embodiment, a neural network processes the performance monitor values to generate operating parameters in real-time.

Abstract: Unavoidable physical phenomena, such as an alpha particle strikes, can cause soft errors in integrated circuits. Materials that emit alpha particles are ubiquitous, and higher energy cosmic particles penetrate the atmosphere and also cause soft errors. Some soft errors have no consequence, but others can cause an integrated circuit to malfunction. In some applications (e.g. driverless cars), proper operation of integrated circuits is critical to human life and safety. To minimize or eliminate the likelihood of a soft error becoming a serious malfunction, detailed assessment of individual potential soft errors and subsequent processor behavior is necessary. Embodiments of the present disclosure facilitate emulating a plurality of different, specific soft errors. Resilience may be assessed over the plurality of soft errors and application code may be advantageously engineered to improve resilience.

Abstract: Unavoidable physical phenomena, such as an alpha particle strikes, can cause soft errors in integrated circuits. Materials that emit alpha particles are ubiquitous, and higher energy cosmic particles penetrate the atmosphere and also cause soft errors. Some soft errors have no consequence, but others can cause an integrated circuit to malfunction. In some applications (e.g. driverless cars), proper operation of integrated circuits is critical to human life and safety. To minimize or eliminate the likelihood of a soft error becoming a serious malfunction, detailed assessment of individual potential soft errors and subsequent processor behavior is necessary. Embodiments of the present disclosure facilitate emulating a plurality of different, specific soft errors. Resilience may be assessed over the plurality of soft errors and application code may be advantageously engineered to improve resilience.

Abstract: Graphics processing unit (GPU) performance and power efficiency is improved using machine learning to tune operating parameters based on performance monitor values and application information. Performance monitor values are processed using machine learning techniques to generate model parameters, which are used by a control unit within the GPU to provide real-time updates to the operating parameters. In one embodiment, a neural network processes the performance monitor values to generate operating parameters in real-time.

Abstract: Integrated circuits, or computer chips, typically include multiple hardware components (e.g. memory, processors, etc.) operating under a shared power (e.g. thermal) constraint that is sourced by one or more power sources for the chip. Typically, the hardware components can be individually configured to operate at certain states (e.g. to operate at a certain frequency by setting a clock speed for a clock dedicated to the hardware component). Thus, each hardware component can be configured to operate at an operating point that is determined to be optimal, usually in terms of achieving some desired goal for a specific application (e.g. frame rates for gaming, etc.). In the context of chip hardware that operates under a shared power/thermal constraint, a method, computer readable medium, and system are provided for determining the optimal operating point for the chip that takes into consideration both performance of the chip and power consumption by the chip.

Abstract: Graphics processing unit (GPU) performance and power efficiency is improved using machine learning to tune operating parameters based on performance monitor values and application information. Performance monitor values are processed using machine learning techniques to generate model parameters, which are used by a control unit within the GPU to provide real-time updates to the operating parameters. In one embodiment, a neural network processes the performance monitor values to generate operating parameters in real-time.

Abstract: An apparatus comprises a dynamic random-access memory (DRAM) for storing data. Refresh control circuitry is provided to control the DRAM to periodically perform a refresh cycle for refreshing the data stored in each memory location of the DRAM. A refresh address sequence generator generates a refresh address sequence of addresses identifying the order in which memory locations of the DRAM are refreshed during the refresh cycle. To deter differential power analysis attacks on secure data stored in the DRAM, the refresh address sequence is generated with the addresses of at least a portion of the memory locations in a random order which varies from refresh cycle to refresh cycle.

Abstract: A storage circuit 2 in the form of a master slave latch includes a slave stage 6 serving as a bit storage circuit. The slave stage 6 includes an inverter chain which when operating in a normal mode includes an even number of inverters 10, 12 and when operating in an random number generation mode includes an odd number of inverters 10, 12, 14 and so functions as a free running ring oscillator. When a switch is made back from the random number generation mode to the normal mode, then the oscillation ceases and a stable pseudo random bit value is output from the bit value storage circuit 6.

Abstract: An integrated circuit is provided with operational mode header transistors which connect a virtual power rail to a VDD power supply. A controller circuit, responsive to a sensed voltage signal from a voltage sensor which reads the virtual rail voltage VVDD, generates a control signal which controls the operational mode transistors. The control signal is derived from an interface voltage power supply that provides higher voltage VDD IO than the VDD power supply and thus able to overdrive the operational mode transistors via either a gate bias voltage or a bulk bias voltage. The amount of leakage through the operational mode transistors is controlled in a closed loop feedback arrangement so as to maintain a predetermined target value or range for the virtual rail voltage. The operational mode transistor may also be controlled to support dynamic voltage and frequency scaling.

Abstract: A semiconductor memory storage device is disclosed. The memory comprises a plurality of storage cells for storing data each storage cell comprising an access control device for providing the storage cell with access to or isolation from a data access port in response to an access control signal, access control circuitry for transmitting the access control signal along an access control line to control a plurality of the access control devices connected to the access control line.

Abstract: A data processing apparatus is provided which comprises a processor unit configured to perform data processing operations in response to a sequence of instructions and a storage unit configured to store data values for access by the processor unit when performing its data processing operations. Redundant error control data is stored in association with the data values, the redundant error control data enabling identification of an error in the data values. The data processing apparatus also comprises a data scrubbing unit configured to perform a data scrubbing process on at least a subset of the data values, the data scrubbing process comprising determining with reference to the redundant error control data if an error is present in that subset of data values and, where possible, correcting that error with reference to the redundant error control data.

Abstract: A storage circuit 2 in the form of a master slave latch includes a slave stage 6 serving as a bit storage circuit. The slave stage 6 includes an inverter chain which when operating in a normal mode includes an even number of inverters 10, 12 and when operating in an random number generation mode includes an odd number of inverters 10, 12, 14 and so functions as a free running ring oscillator. When a switch is made back from the random number generation mode to the normal mode, then the oscillation ceases and a stable pseudo random bit value is output from the bit value storage circuit 6.

Abstract: A latching device includes input and output latching elements to receive and output data values wherein the input and output elements are configured to receive a first and second clocks, respectively. The clocks have the same frequency but are inverted. The elements are transparent and transmit data between an input and an output in response to the first value of a received clock and are opaque and hold the data value in response to a second value of the received clock, such that in response to the first and second clocks the input data value is clocked through the input and output elements to the output. The device includes a device for selecting an operational data value or a diagnostic data value for input to the input element in response to a value of a diagnostic enable signal indicating a functional mode or a diagnostic mode.

Abstract: A data processing apparatus is provided comprising data processing circuitry configured to perform data processing operations. A plurality of state retention circuits forms part of the data processing circuitry and these circuits are configured to hold respective state values at respective nodes of the data processing circuitry it enters a low power mode. One or more scan paths connect the plurality of state retention circuits together in series, such that the state values may be scanned into and out of the respective nodes. A plurality of parity information generation elements are coupled to the scan path(s) and configured to generate parity information indicative of the respective state values held at those respective nodes by the state retention circuits.