Abstract: The technology generally relates to high bandwidth memory (HBM) and optical connectivity stacking. Disclosed systems and methods herein allow for 3D-stacking of HBM dies that are interconnected with an optical interface in a manner that allows for compact, high-performance computing. An optical chiplet can be configured to be placed onto a stack of HBM dies, with a cooling die that is positioned between the HBM dies and the optical chiplet. The optical chiplet may be configured to connect the HBM optics module package to one or more other components of the package via to one or more optical fibers.

Abstract: A microelectronic system may include a substrate having a first surface, one or more interposers mounted to and electrically connected to the first surface, first and second application specific integrated circuits (ASICs) each at least partially overlying and electrically connected to one of the interposers, a plurality of high-bandwidth memory elements (HBMs) each at least partially overlying and electrically connected to one of the interposers, and an active silicon bridge mounted to and electrically connected to the first surface and providing an electrical connection between the first and second ASICs, the active silicon bridge having active microelectronic devices therein. The microelectronic system may be configured such that the first and second ASICs and the active silicon bridge each have a purely digital CMOS interface therein. A plurality of bumps providing the electrical connection between the ASICs and the active silicon bridge may be configured to receive serial data therethrough.

Abstract: The technology generally relates to disaggregating memory from an application specific integrated circuit (“ASIC”) package. For example, a high-bandwidth memory (“HBM”) optics module package may be connected to an ASIC package via one or more optical links. The HBM optics module package may include HBM dies(s), HBM chiplet(s) and an optical chiplet. The optical chiplet may be configured to connect the HBM optics module to one or more optical fibers that form an optical link with one or more other components of the ASIC package. By including an optical chiplet in the HBM optics module package, the HBM optics module package may be disaggregated from an ASIC package.

Abstract: A system contains a machine learning application specific integrated circuit (ASIC) and a power supply unit. The power supply unit and the ASIC are configured to be in data communication through dedicated pins on the ASIC and the power supply unit. The power supply unit detects a present power consumption of the ASIC. Upon determining that a threshold condition has been met, the power supply unit, responsive to the condition sends a digital signal to the ASIC. The ASIC contains a synchronizer which synchronizes the digital signal to be consistent with the ASICs internal clock frequency. A chip manager the synchronized signal and other signals to generate a throttling mask. The throttling mask is sent to a sequencer of the ASIC, which then limits the instruction flow into the processing units of the ASIC based on the mask. This in turn limits the power being consumed by the ASIC.

Abstract: A system contains a machine learning application specific integrated circuit (ASIC) and a power supply unit. The power supply unit and the ASIC are configured to be in data communication through dedicated pins on the ASIC and the power supply unit. The power supply unit detects a present power consumption of the ASIC. Upon determining that a threshold condition has been met, the power supply unit, responsive to the condition sends a digital signal to the ASIC. The ASIC contains a synchronizer which synchronizes the digital signal to be consistent with the ASICs internal clock frequency. A chip manager the synchronized signal and other signals to generate a throttling mask. The throttling mask is sent to a sequencer of the ASIC, which then limits the instruction flow into the processing units of the ASIC based on the mask. This in turn limits the power being consumed by the ASIC.

Abstract: A method and system for controlling performance of a workload partitioned among a plurality of accelerator chips of a multi-chip system. One or more processors may receive performance speed data for each of the accelerator chips, obtain a model of the partitioned workload, determine a portion of the workload that is either overworked or underworked based on the model of the partitioned workload and the performance speed data for each of the plurality of accelerator chips, and adjust a performance speed of an accelerator chip that performs the portion of the partitioned workload that is either overworked or underworked.

Abstract: A system contains a machine learning application specific integrated circuit (ASIC) and a power supply unit. The power supply unit and the ASIC are configured to be in data communication through dedicated pins on the ASIC and the power supply unit. The power supply unit detects a present power consumption of the ASIC. Upon determining that a threshold condition has been met, the power supply unit, responsive to the condition sends a digital signal to the ASIC. The ASIC contains a synchronizer which synchronizes the digital signal to be consistent with the ASICs internal clock frequency. A chip manager the synchronized signal and other signals to generate a throttling mask. The throttling mask is sent to a sequencer of the ASIC, which then limits the instruction flow into the processing units of the ASIC based on the mask. This in turn limits the power being consumed by the ASIC.

Abstract: An embodiment includes a circuit block configured to distribute a power signal to a plurality of voltage sense signals, and a voltage regulator configured to generate a regulated voltage level on the power signal. The embodiment also includes a sensing circuit configured to perform a sequence of comparisons of respective voltage levels of the plurality of voltage sense signals using a selection criterion. To perform the sequence of comparisons, the sensing circuit may be configured to select either a first voltage sense signal or a second voltage sense signal to generate a first output voltage sense signal. The sensing circuit may also be configured to select either a third voltage sense signal or a previously generated output voltage sense signal to generate a feedback signal. The voltage regulator circuit may be further configured to modify the regulated voltage level using the feedback signal.

Abstract: A processing device can identify gates of an integrated circuit design having a slack value less than a predefined slack threshold. The processing device can further, for each of the identified gates, determine (i) a number of nodes of the integrated circuit design that experience a timing slack improvement if the gate is swapped with an alternative implementation having a reduced delay or (ii) a sum of timing slack improvements experienced by nodes of the integrated circuit design if the gate is swapped with the alternative implementation having a reduced delay. The processing device can still further swap the gate with the alternative implementation having the reduced delay if the determined number or sum is greater than a corresponding predetermined threshold.

Abstract: Disclosed is a digital voltage regulator system and method for mitigating voltage droop in an integrated circuit. If an unacceptable voltage droop is detected, the digital voltage regulator may take action to allow the power supply voltage to recover. A digital voltage regulator in accordance with embodiments discussed herein detects voltage droop by comparing a power supply voltage measurement with a threshold voltage. The threshold voltage may be calibrated based on power supply voltage measurements taken while the integrated circuit is operating.

Abstract: A method for determining flop circuit types includes performing a layout of an IC design including arranging master and slave latches of each of a plurality of flops to receive first and second clock signals, respectively. The initial IC design may then be implemented (e.g., on a silicon substrate). After implementation, the IC may be operated in first and second modes. In the first mode, the master latch of each flop is coupled to receive a first clock signal. In the second mode, the first clock signal is inhibited and the master latch is held transparent. The slave latch of each flop operates according to a second clock signal in both the first and second modes. The method further includes determining, for each flop, whether that flop is to operate as a master-slave flip-flop or as a pulse flop in a subsequent revision of the IC.

Abstract: A method for determining flop circuit types includes performing a layout of an IC design including arranging master and slave latches of each of a plurality of flops to receive first and second clock signals, respectively. The initial IC design may then be implemented (e.g., on a silicon substrate). After implementation, the IC may be operated in first and second modes. In the first mode, the master latch of each flop is coupled to receive a first clock signal. In the second mode, the first clock signal is inhibited and the master latch is held transparent. The slave latch of each flop operates according to a second clock signal in both the first and second modes. The method further includes determining, for each flop, whether that flop is to operate as a master-slave flip-flop or as a pulse flop in a subsequent revision of the IC.

Abstract: A system for selecting gates for an integrated circuit design may include at least one processing device configured to identify gates of the integrated circuit design having a slack value less than a predefined slack threshold. The at least one processing device may be further configured to, for each of the identified gates, determine (i) a number of nodes of the integrated circuit design that experience a timing slack improvement if the gate is swapped with an alternative implementation having a reduced delay or (ii) a sum of timing slack improvements experienced by nodes of the integrated circuit design if the gate is swapped with the alternative implementation having a reduced delay. The at least one processing device may still be further configured to swap the gate with the alternative implementation having the reduced delay if the determined number or sum is greater than a corresponding predetermined threshold.

Abstract: Disclosed is a digital voltage regulator system and method for mitigating voltage droop in an integrated circuit. If an unacceptable voltage droop is detected, the digital voltage regulator may take action to allow the power supply voltage to recover. A digital voltage regulator in accordance with embodiments discussed herein detects voltage droop by comparing a power supply voltage measurement with a threshold voltage. The threshold voltage may be calibrated based on power supply voltage measurements taken while the integrated circuit is operating.

Abstract: A method for creating a wire load model using specific interconnect configuration information is provided. Further, a program that creates a wire load model by curve-interconnect fitting parasitic information and interconnect configuration information is provided. Further, a computer system capable of creating an accurate wire load model using parasitic information specific to particular metal layers is provided.

Abstract: An automated apparatus detects electromigration violations in an integrated circuit design. Starting from the lowest hierarchy of the design so far completed, the parasitic (resistance and capacitance) component values extracted from a layout file are propagated up. Then, at the top-most level, lumping algorithms are employed to calculate the parasitic values for all of the top-most level nets. These values are then passed back down to the lower levels and then at each level, the layout is checked using previously computed parasitic values and EM limits. A peak current, AC-average current and AC-rms current are calculated for every layout, and then compared with the process EM rules for violations, in which the optimum line width and number of vias are specified for each interconnection.

Abstract: An automated method detects electromigration violations in an integrated circuit design. Starting from the lowest hierarchy of the design so far completed, the parasitic (resistance and capacitance) component values extracted from a layout file are propagated up. Then, at the top-most level, lumping algorithms are employed to calculate the parasitic values for all of the top-most level nets. These values are then passed back down to the lower levels and then at each level, the layout is checked using previously computed parasitic values and EM limits. A peak current, AC-average current and AC-rms current are calculated for every layout, and then compared with the process EM rules for violations, in which the optimum line width and number of vias are specified for each interconnection.