Abstract: Described herein are fabrication processes and resulting transistor arrangements with trench contacts that have two parts—a first trench contact (TCN1) and a second trench contact (TCN2)—stacked over one another, and with gate contacts (VCGs). In such transistor arrangements, the TCN1 may be self-aligned to adjacent gates and may be used to make cell-level connections, the TCN2 may also make cell-level connections and may be provided after the self-aligned TCN1 formation and may have an inverse taper shape, the spacer around the TCN2 may be a higher dielectric constant dielectric material than conventional spacer materials, and the VCGs may be formed without the presence of any gate caps or after using only thin temporary gate caps. Fabrication processes and transistor arrangement described herein may provide several improvements in terms of increased edge placement error margin, cost-efficiency, and device performance.

Abstract: A stiffener for an integrated circuit (IC) package assembly including an IC die electrically interconnected to a substrate. The stiffener is to be mechanically attached to the substrate adjacent to at least one edge of the IC die and have a coefficient of linear thermal expansion (CTE) exceeding that of the substrate. The stiffener may be an “anti-invar” metallic alloy. Anti-invar alloys display “anti-invar” behavior where thermal expansion of the material is significantly enhanced relative to other compositions of the particular alloy system. A package stiffener may be a high-Mn steel, for example, such as ASTM International A128. In other examples, a package stiffener is a MnCuNi, FeNiMn, or FeNiCr alloy having an average CTE over a range of 25-100° C. of at least 18 ppm, and a room temperature modulus of elasticity of at least 120 GPa.

Abstract: Various embodiments are generally directed to techniques to tune a scale parameter for activations in binary neural networks, such as based on estimating a gradient for the scale parameter using quantization error, for instance. Some embodiments are particularly directed to tuning the scale parameter for activations by estimating the gradient for the scale parameter using a first “force” based on quantization error and a second, opposing, “force” based on clipping error. For instance, the first “force” based on the quantization error may give a gradient for the scale parameter that pushes the scale parameter lower to reduce the quantization error and the second “force” based on the clipping error may give a gradient for the scale parameter that moves the scale parameter higher to reduce the number of activations that are higher than a current scale parameter.

Abstract: Embodiments described herein are generally directed to maintaining contiguity of virtual to physical address mappings to exploit a contiguity-aware TLB. In an example, information regarding a migration set of one or more pages within a physical address space that have been identified for migration from a source tier of memory to a target tier of memory is received in which the physical address space comprises a first contiguous region of physical memory addresses and a VMA includes a second contiguous region of virtual memory addresses corresponding to the first contiguous region. It is determined whether the migration would break contiguity of a mapping maintained by a contiguity-aware TLB between pages of the first contiguous region and pages of the second contiguous region. Responsive an affirmative determination, discontinuities within the mapping resulting from the migration are minimized by intelligently increasing or decreasing the migration set.

Abstract: An apparatus to facilitate generating a memory bandwidth stack for visualizing memory bandwidth utilization is disclosed. The apparatus includes processors to receive data corresponding to a memory cycle occurring during a total execution time of an application executed by the one or more processors; for the memory cycle, assign the memory cycle to a component of a bandwidth stack based on analysis of the data and in accordance with a prioritization scheme; for the component, determine a portion of the bandwidth stack to account to the component based at least in part on the assignment of the memory cycle to the component; and generate the bandwidth stack by at least representing the portion accounted to the component in the bandwidth stack.

Abstract: Techniques are described for providing consistent memory operations and security across electronic circuitry components having disparate memory and/or security architectures when integrating such disparately architected components within a single system, such as a system on chip. A programmable logical hierarchy of isolated memory region (IMR) enforcement circuits is provided to protect such IMRs, allowing or preventing memory access requests from one of multiple distinct circuitry components based on configuration registers for the IMR enforcement circuits. Integration of multiple trust domain architectures associated with the multiple distinct circuitry components is facilitated via trust domain conversion bridge circuitry that includes translation logic for generating information in accordance with a first trust domain architecture based on information provided in accordance with a distinct second trust domain architecture.

Abstract: An inductor can be formed in a coreless electronic substrate from magnetic materials and/or fabrication processes that do not result in the magnetic materials leaching into plating and/or etching solutions/chemistries, and results in a unique inductor structure. This may be achieved by forming the inductors from magnetic ferrites. The formation of the electronic substrates may also include process sequences that prevent exposure of the magnetic ferrites to the plating and/or etching solutions/chemistries.

Abstract: Various embodiments are generally directed to virtualized systems. A first guest memory page may be identified based at least in part on a number of accesses to a page table entry for the first guest memory page in a page table by an application executing in a virtual machine (VM) on the processor, the first guest memory page corresponding to a first byte-addressable memory. The execution of the VM and the application on the processor may be paused. The first guest memory page may be migrated to a target memory page in a second byte-addressable memory, the target memory page comprising one of a target host memory page and a target guest memory page, the second byte-addressable memory having an access speed faster than an access speed of the first byte-addressable memory.

Abstract: A method is provided for using an integrated circuit design tool to optimize a circuit design for an integrated circuit. Optimizations that affect first circuits in the circuit design and that are performed during synthesis of the circuit design for the integrated circuit are recorded in first records in a database. Second records are recorded in the database that indicate second circuits in the circuit design that fanin to or fanout from the first circuits and that are affected by the optimizations. A root cause of at least one of the optimizations is determined using the first and the second records in the database. A sequence of the optimizations affecting the first and the second circuits is determined using the first and the second records in the database.

Abstract: Apparatus and method for correcting image regions following upsampling or frame interpolation. For example, one embodiment of an apparatus comprises a machine-learning engine to evaluate at least a first image in a sequence of images generated by a real-time interactive application, the machine learning engine to responsively use previously learned data to generate an upsampled or interpolated image comprising a plurality of pixel patches. In one embodiment, each pixel patch is associated with a confidence value reflecting how accurately the pixel patch was generated by the machine learning engine. A selective ray tracing engine identifies a first pixel patch to be corrected based a first confidence value corresponding to the first pixel patch being lower than a threshold and performs ray tracing operations on a first portion of the first image to generate a corrected first pixel patch.

Abstract: Disclosed herein are integrated circuit (IC) packages, antenna boards, antenna modules, and communication devices (e.g., for millimeter wave communications). For example, in some embodiments, an antenna module may include: a logic die; a radio frequency front-end (RFFE) die in electrical communication with the logic die; and an antenna patch, wherein the RFFE die is closer to the antenna patch than the logic die is to the antenna patch.

Abstract: Techniques to enable dynamic bandwidth management at the physical layer level while maintaining backwards compatibility in wireless systems is provided. Furthermore, techniques for reducing the occurrence of exposed nodes are provided. A transmitter may transmit a frame including an indication that a PHY layer sub-header defining a bandwidth associated with a channel is present. Furthermore, the transmitter may transmit a third frame after receiving a second frame from a receiver to indicate to legacy stations that the TXOP was successful.

Abstract: There is disclosed in one example a computing system, including: a processor; a memory; and a memory encryption engine (MEE) including circuitry and logic to: allocate a protected isolated memory region (IMR); encrypt the protected IMR; set an access control policy to allow access to the IMR by a device identified by a device identifier; and upon receiving a memory access request directed to the IMR, enforce the access control policy.

Abstract: A platform comprising numerous reconfigurable circuit components arranged to operate as primary and redundant circuits is provided. The platform further comprises security circuitry arranged to monitor the primary circuit for anomalies and reconfigurable circuit arranged to disconnect the primary circuit from a bus responsive to detection of an anomaly. Furthermore, the present disclosure provides for the quarantine, refurbishment and designation as redundant, the anomalous circuit.

Abstract: Tunneling Field Effect Transistors (TFETs) are promising devices in that they promise significant performance increase and energy consumption decrease due to a steeper subthreshold slope (for example, smaller sub-threshold swing). In various embodiments, vertical fin-based TFETs can be fabricated in trenches, for example, silicon trenches. In another embodiment, vertical TFETs can be used on different material systems acting as a substrate and/or trenches (for example, Si, Ge, III-V semiconductors, GaN, and the like). In one embodiment, the tunneling direction in the channel of the vertical TFET can be perpendicular to the Si substrates. In one embodiment, this can be different than the tunneling direction in the channel of lateral TFETs.

Abstract: Microelectronic assemblies, and related devices and methods, are disclosed herein. For example, in some embodiments, a microelectronic assembly may include a package substrate having a core substrate with a first conductive structure having a first thickness on the core substrate, and a second conductive structure having a second thickness on the core substrate, where the first thickness is different than the second thickness.

Abstract: Systems, methods, computer program products, and apparatuses for scene retrieval are provided. Images can be captured with a depth camera and image data encoded with both color and depth indications. A convolutional neural network comprising a fast channel wide block is provided. Image descriptors can be extracted from the images based on output from the fast channel wide block. Such image descriptors can be used to retrieve scenes from a SLAM process for purposes of localization.

Abstract: An integrated circuit includes sectors of logic circuits. Each of the sectors of logic circuits includes a local sector manager controller circuit that provides an indication to perform a memory test. Each of the sectors also includes a built-in-self-test circuit that is configurable to generate a control signal and expected data in response to the indication to perform the memory test. Each of the sectors also includes a memory circuit that outputs read data in response to the control signal during the memory test. Each of the sectors further includes a comparator circuit configurable to compare the read data with the expected data during the memory test to generate a test result that is provided to the built-in-self-test circuit.

Abstract: Microelectronic assemblies, and related devices and methods, are disclosed herein. For example, in some embodiments, a microelectronic assembly may a die having a front side and a back side, the die comprising a first material and conductive contacts at the front side; and a thermal layer attached to the back side of the die, the thermal layer comprising a second material and a conductive pathway, wherein the conductive pathway extends from a front side of the thermal layer to a back side of the thermal layer.

Abstract: Thermally compensated waveguides are disclosed herein. According to one aspect, the present disclosure proposes new ways to combine negative TOC (NTOC) material layers within the waveguides. NTOC materials can be implemented in one or more of a cladding layer, a core rib/channel waveguide, a horizontally segmented waveguide, a vertically segmented waveguide, a sub-wavelength grating structure, and/or in various other waveguide structure implementations including arbitrary core or cladding shapes. The integration of NTOC materials improves the temperature dependence of the waveguide spectrum. The need for fast and efficient optical-based technologies is increasing as Internet data traffic growth rate is overtaking voice traffic, pushing the need for optical communications.

Abstract: One embodiment provides a graphics processor comprising a system interface and circuitry coupled with the system interface. The circuitry includes an execution resource and a preemption status register. The execution resource is configured to execute an instruction. During execution of the instruction, the execution resource is to receive a request to preempt execution of a thread associated with the instruction and, based on a value stored in the preemption status register, execute at least one additional instruction after receipt of the request to preempt execution of the thread.

Abstract: Systems, apparatus, methods, and techniques for reporting an attack or intrusion into an in-vehicle network are provided. The attack can be broadcast to connected vehicles over a vehicle-to-vehicle network. The broadcast can include an indication of a sub-system involved in the attack and can include a request for assistance in recovering from the attack. Connected vehicles can broadcast responses over the vehicle-to-vehicle network. The responses can include indications of data related to the compromised sub-system. The vehicle can receive the responses and can use the responses to recover from the attack, such as, estimate data.

Abstract: Techniques related to generating holographic images are discussed. Such techniques include application of a machine learning model to the target image to generate data that is used to enable the determination of a phase pattern via an iterative propagation feedback model. The iterative propagation feedback model is used to generate a feedback strength value, which is then used to generate a phase diffraction pattern for presentation at a holographic plane.

Abstract: Embodiments are directed to a session management framework for secure communications between host systems and trusted devices. An embodiment of computer-readable storage mediums includes instructions for establishing a security agreement between a host system and a trusted device, the host device including a trusted execution environment (TEE); initiating a key exchange between the host system and the trusted device, including sending a key agreement message from the host system to the trusted device; sending an initialization message to the trusted device; validating capabilities of the trusted device for a secure communication session between the host system and the trusted device; provisioning secrets to the trusted device and initializing cryptographic parameters with the trusted device; and sending an activate session message to the trusted device to activate the secure communication session over a secure communication channel.

Abstract: Systems, methods, computer program products, and apparatuses for low latency, fully reconfigurable hardware logic for ensemble classification methods, such as random forests. An apparatus may comprise circuitry for an interconnect and circuitry for a random forest implemented in hardware. The random forest comprising a plurality of decision trees connected via the interconnect, each decision tree comprising a plurality of nodes connected via the interconnect. A first decision tree of the plurality of decision trees comprising a first node of the plurality of nodes to: receive a plurality of elements of feature data via the interconnect, select a first element of feature data, of the plurality of elements of feature data, based on a configuration of the first node, and generate an output based on the first element of feature data, an operation, and a reference value, the operation and reference value specified in the configuration of the first node.

Abstract: Disclosed herein are integrated circuit (IC) supports with microstrips, and related embodiments. For example, an IC support may include a first microstrip; a first surface dielectric region over the first microstrip, wherein the first surface dielectric region has a first thickness, and the first thickness is nonzero; a second microstrip; and a second surface dielectric region over the second microstrip, wherein the second surface dielectric region has a second thickness, the second thickness is nonzero, and the first thickness is different than the second thickness.

Abstract: An integrated circuit includes a multiplexer circuit coupled to receive a first clock signal and a second clock signal and coupled to provide an output clock signal to a channel. A protection circuit is coupled to receive a feedback signal from the channel. The protection circuit causes the multiplexer circuit to provide oscillations in the second clock signal to the output clock signal in response to the feedback signal indicating that the channel is idle to cause the channel to be in a protection mode that reduces degradation from bias temperature instability. The protection circuit causes the multiplexer circuit to provide oscillations in the first clock signal to the output clock signal in response to the feedback signal indicating that the channel is active.

Abstract: An apparatus is provided which comprises: a substrate, the substrate comprising crystalline material, a first set of one or more contacts on a first substrate surface, a second set of one or more contacts on a second substrate surface, the second substrate surface opposite the first substrate surface, a first via through the substrate coupled with a first one of the first set of contacts and with a first one of the second set of contacts; a second via through the substrate coupled with a second one of the first set of contacts and with a second one of the second set of contacts, a trench in the substrate from the first substrate surface toward the second substrate surface, wherein the trench is apart from, and between, the first via and the second via, and dielectric material filling the trench. Other embodiments are also disclosed and claimed.

Abstract: Techniques for controlling current through memory cells is disclosed. In the illustrative embodiment, a fine-grained current source and a coarse-grained current source can both be activated to perform an operation on a phase-change memory cell. The coarse-grained current source is briefly activated to charge up the capacitance of an electrical path through the memory cell and then turned off. The fine-grained current source applies a current pulse to perform the operation on the memory cell, such as a reset operation. By charging up the electrical path quickly with the coarse-grained current source, the fine-grained current source can quickly perform the operation on the memory cell, reducing the thermal disturbance caused by the operation on nearby memory cells.

Abstract: Disclosed herein are microelectronic structures including bridges, as well as related assemblies and methods. In some embodiments, a microelectronic structure may include a substrate and a bridge.

Abstract: A graphics processor is described that includes a single instruction, multiple thread (SIMT) architecture including hardware multithreading. The multiprocessor can execute parallel threads of instructions associated with a command stream, where the multiprocessor includes a set of functional units to execute at least one of the parallel threads of the instructions. The set of functional units can include a mixed precision tensor processor to perform tensor computations. The functional units can also include circuitry to analyze statistics for output values of the tensor computations, determine a target format to convert the output values, the target format determined based on the statistics for the output values and a precision associated with a second layer of the neural network, and convert the output values to the target format.

Abstract: Techniques related to training and implementing a bidirectional pairing architecture for object detection are discussed. Such techniques include generating a first enhanced feature map for each frame of a video sequence by processing the frames in a first direction, generating a second enhanced feature map for frame by processing the frames in a second direction opposite the first, and determining object detection information for each frame using the first and second enhanced feature map for the frame.

Abstract: An interface couples a controller to a physical layer (PHY) block, where the interface includes a set of data pins comprising transmit data pins to send data to the PHY block and receive data pins to receive data from the PHY block. The interface further includes a particular set of pins to implement a message bus interface, where the controller is to send a write command to the PHY block over the message bus interface to write a value to at least one particular bit of a PHY message bus register, bits of the PHY message bus register are mapped to a set of control and status signals, and the particular bit is mapped to a recalibration request signal to request that the PHY block perform a recalibration.

Abstract: Apparatus and method for efficient BVH construction. For example, one embodiment of an apparatus comprises: a memory to store graphics data for a scene including a plurality of primitives in a scene at a first precision; a geometry quantizer to read vertices of the primitives at the first precision and to adaptively quantize the vertices of the primitives to a second precision associated with a first local coordinate grid of a first BVH node positioned within a global coordinate grid, the second precision lower than the first precision; a BVH builder to determine coordinates of child nodes of the first BVH node by performing non-spatial-split binning or spatial-split binning for the first BVH node using primitives associated with the first BVH node, the BVH builder to determine final coordinates for the child nodes based, at least in part, on an evaluation of surface areas of different bounding boxes generated for each of the child node.

Abstract: In one example, a system for asymmetric device attestation includes a physically unclonable function (PUF) configured to generate a response to a challenge. A pseudo-random number generator generates a set of random numbers based on the response. A key generator determines co-prime numbers in the set of random numbers and generates a key pair using the co-prime numbers, wherein the public key is released to a manufacturer of the component for attestation of authenticity of the component. Through extending the PUF circuitry with a pseudo-random number generator, the present techniques are able to withstand unskilled and skilled hardware attacks, as the secret derived from the PUF is immune to extraction.

Abstract: Systems and methods for improving cache efficiency and utilization are disclosed. In one embodiment, a graphics processor includes processing resources to perform graphics operations and a cache controller of a cache coupled to the processing resources. The cache controller is configured to control cache priority by determining whether default settings or an instruction will control cache operations for the cache.

Abstract: An DNN accelerator includes a column of PEs and an external adder assembly for performing depthwise convolution. Each PE includes register files, multipliers, and an internal adder assembly. Each register file can store an operand (input operand, weight operand, etc.) of the depthwise convolution. The operand includes a sequence of elements, each of which corresponds to a different depthwise channel. A multiplier can perform a sequence of multiplications on two operands, e.g., an input operand and a weight operand, and generate a product operand. The internal adder assembly can accumulate product operands and generate an output operand of the PE. The output operand includes output elements, each of which corresponds to a different depthwise channel. The operands may be reused in different rounds of operations by the multipliers. The external adder assembly can accumulate output operands of multiple PEs and generate an output operand of the PE column.

Abstract: Methods, systems and apparatuses may provide for technology that obtains multi-camera video data including a first 2D image corresponding to a first camera and a second 2D image corresponding to a second camera. The technology may also identify an association between a first instance of a 3D object in the first 2D image and a second instance of the 3D object in the second 2D image, and automatically generate a 3D bounding box around the 3D object based on the association between the first instance and the second instance.

Abstract: Systems, apparatuses and methods may include a source device that generates a scene change notification in response to a movement of a camera, modifies an encoding scheme associated with the video content captured by the camera in response to the scene change notification, identifies a full-frame difference threshold, wherein scene analysis information includes frame difference data, and compares the frame difference data to an intermediate threshold that is less than the full-frame difference threshold, wherein the scene change notification is generated when the frame difference data exceeds the intermediate threshold. A sink device may obtain transport quality data associated with video content, modify an output parameter of a display based on the transport quality data, determine a view perspective of a still image containing a plurality of image slices, retrieve only a subset of the plurality of image slices based on the view perspective and decode the retrieved subset.

Abstract: A memory cell can include a top lamina layer, a bottom lamina layer, and a phase change material (PCM) layer between the top lamina layer and the bottom lamina layer. The PCM layer can have a top surface in direct contact with the top lamina layer and a bottom surface in direct contact with the bottom lamina layer. The top surface of the PCM layer and the bottom surface of the PCM layer can have a structurally stabilizing width ratio.

Abstract: A method comprises initializing, by an accelerator device of the computing device, an authentication tag in response to an initialization command from a trusted execution environment of the computing device, initiating a transfer, by the accelerator device, of data between a host memory and an accelerator device memory in response to a descriptor from the trusted execution environment, wherein the descriptor comprises a target memory address and is indicative of a transfer direction, comparing, in a memory range selection engine comprising at least one comparator to compare the target memory address with a plurality of address ranges and select a cryptographic key from the plurality of plurality of address range registers based on the target memory address, performing, by the accelerator device, a cryptographic operation with the data in response to transferring the data, updating, by the accelerator device, the authentication tag in response to transferring the data, and finalizing, by the accelerator device

Abstract: An apparatus to facilitate graphics rendering is disclosed. The apparatus comprises sequencer hardware to operate in a tile mode to render objects, including performing batch formation to generate one or more batches of received objects, performing tile sequencing for each of the objects to compute tile fill intersects for each of the objects and performing a play sequencing of each of the objects.

Abstract: Embodiments are generally directed to thread group scheduling for graphics processing. An embodiment of an apparatus includes a plurality of processors including a plurality of graphics processors to process data; a memory; and one or more caches for storage of data for the plurality of graphics processors, wherein the one or more processors are to schedule a plurality of groups of threads for processing by the plurality of graphics processors, the scheduling of the plurality of groups of threads including the plurality of processors to apply a bias for scheduling the plurality of groups of threads according to a cache locality for the one or more caches.

Abstract: A data center cluster includes a plurality of host systems coupled to a network processing device by a Compute Express Link (CXL) switch, where the network processing device includes memory to implement a memory pool for the data center cluster. Request and responses are communicated within the data center cluster using the memory pool and the network processing device manages communication within the data center cluster.

Abstract: Disclosed herein are selector devices and related devices and techniques. In some embodiments, a selector device may include a first electrode, a second electrode, and a selector material between the first electrode and the second electrode. The selector material may include germanium, tellurium, and sulfur.

Abstract: Method and system for performing data movement operations is described herein. One embodiment of a method includes: storing data for a first memory address in a cache line of a memory of a first processing unit, the cache line associated with a coherency state indicating that the memory has sole ownership of the cache line; decoding an instruction for execution by a second processing unit, the instruction comprising a source data operand specifying the first memory address and a destination operand specifying a memory location in the second processing unit; and responsive to executing the decoded instruction, copying data from the cache line of the memory of the first processing unit as identified by the first memory address, to the memory location of the second processing unit, wherein responsive to the copy, the cache line is to remain in the memory and the coherency state is to remain unchanged.

Abstract: A mechanism where the locked pages are saved and restored by a hardware accelerator which is transparent to the OS. Prior to standby entry, the OS puts all DMA capable devices in the lowest-powered device low-power state after disabling bus mastering. The OS flushes all pageable memory to an NVM (in segments that are kept in self-refresh) and provides a list of pinned and locked pages in the DRAM to a power management controller (p-unit). The p-unit checks for all Bus Mastering DMA to be turned off and checks if a next OS timer wake event (TNTE) is greater than a threshold, to decide whether to enable or disable PASR or MPSM in Standby. If the conditions are met, the p-unit triggers a hardware accelerator to consolidate the pinned and locked pages in the DRAM to certain segment(s) of the DRAM during standby states, making it transparent to the OS.

Abstract: Methods and apparatus relating to techniques for region-based deterministic memory safety are described. In some embodiment, one or more instructions may be used to encrypt, decrypt, and/or check a pointer to a portion of the data stored in memory. The portion of the data is stored in a first region of the memory. The first region of the memory includes a plurality of identically sized allocation slots. Other embodiments are also disclosed and claimed.

Abstract: An apparatus to facilitate compute optimization is disclosed. The apparatus includes a mixed precision core including mixed-precision execution circuitry to execute one or more of the mixed-precision instructions to perform a mixed-precision dot-product operation comprising to perform a set of multiply and accumulate operations.

Abstract: Disclosed herein are IC structures with decoupling capacitors based on dummy TSV plates provided in a support structure (e.g., a substrate, a die, a wafer, or a chip). An example decoupling capacitor includes first and second capacitor plates and a capacitor insulator between them. Each capacitor plate is a different blind, plate-like opening in the support structure, the openings at least partially filled with one or more conductive materials. The capacitor plate openings are referred to herein as “dummy TSV plates” because they may be fabricated while providing regular TSV openings in the support structure. Such decoupling capacitors may be better suited for high-speed microprocessor applications than conventional off-chip decoupling capacitors and may advantageously allow integrating on-chip decoupling capacitors with an ample amount of capacitive decoupling, limited or no additional processing steps on top of regular TSV processing, and in areas that may not have been used otherwise.