Abstract: Aspects of the disclosure relate generally to the design of the functional units of a processor core, and more specifically, to adding logical operations of a first functional unit of a processor core to a second functional unit of the processor core. In an aspect, a processor core includes a first functional unit configured to provide first functionality, wherein the first functional unit includes circuitry configured to perform a first set of logical operations, a second functional unit configured to provide second functionality different from the first functionality, wherein the second functional unit includes circuitry configured to perform a subset of logical operations of the first set of logical operations, and a data bus connecting the first functional unit and the second functional unit.

Abstract: Cache memory systems employing multiple-level hierarchy cache coherency architecture, and related methods and computer-readable media. A processor-based system includes separate dies that each have a processor and local cache memory logically forming a portion of global cache memory for a system address space. To provide a single point of cache coherency in the global cache memory, the processor-based system includes a proxy cache controller circuit in each die, and a global cache controller circuit. The global cache controller circuit can communicate with the proxy cache controller circuits to maintain single point of cache coherency in the global cache memory. Thus, a cache coherency protocol based on a single point of cache coherency can be implemented.

Abstract: Apparatuses, systems, and methods for controlling cache allocations in a configurable combined private and shared cache in a processor-based system. The processor-based system is configured to receive a cache allocation request to allocate a line in a share cache structure, which may further include a client identification (ID). The cache allocation request and the client ID can be compared to a sub-non-uniform memory access (NUMA) (sub-NUMA) bit mask and a client allocation bit mask to generate a cache allocation vector. The sub-NUMA bit mask may have been programmed to indicate that processing cores associated with a sub-NUMA region are available, whereas processing cores associated with other sub-NUMA regions are not available, and the client allocation bit mask may have been programmed to indicate that processing cores are available.

Abstract: Disclosed are techniques for a processing device including a mesh network connecting at least a request node device, multiple home node devices, and multiple slave node devices. In an aspect, the request node device may select a target home node device. The home node devices may be divided into M groups of home node devices. The request may be routed from the request node device to the target home node device. The target home node device may select a target slave node device from a target group of M groups of slave node devices associated with a target group of the M groups of home node devices to which the target home node device belongs. The request may be routed from the target home node device to the target slave node device.

Abstract: Aspects disclosed in the detailed description include performing snoop filter replacement based on history-augmented victimization priority values of snoop filter entries in processor-based devices. In an exemplary aspect, a Fully Coherent Home Node (HN-F) circuit of a processor-based device receives, from a Fully Coherent Request Node (RN-F) circuit, a transaction request comprising a memory address. The HN-F circuit determines a victimization priority value based on the transaction request. Upon determining that no snoop filter entry in a snoop filter of the HN-F circuit stores the memory address and determining that no snoop filter entries are available for allocation, the HN-F circuit selects a target snoop filter entry that stores a highest victimization priority value among the snoop filter entries, writes the current memory address of the target snoop filter back to memory, and then stores the memory address and the victimization priority value in the target snoop filter entry.

Abstract: Disclosed are techniques for memory resource control using Memory System Resource Partitioning and Monitoring (MPAM). In an aspect, a method of memory-system resource usage monitoring on a processing unit may include attaching a partition identifier from a set of partition identifiers to each memory access request of a plurality of memory access requests on an interconnect. The method may also include interleaving each memory access request of the plurality of memory access requests to a set of memory system components. The method may also include determining a first bandwidth associated with a first memory system component of the set of memory system components. The method may also include applying the first bandwidth associated with the first memory system component to one or more other memory system components of the set of memory system components based at least in part on the interleaving each memory access request.

Abstract: Apparatus and methods for extending functionality of memory controllers using a loopback mode for testing are disclosed herein. In one aspect, a processor-based device provides a memory access intercept circuit configured to receive a memory write request that is directed to and received by a memory controller. The memory access intercept circuit transmits proxy write data to the memory controller, and intercepts write data directed to the memory controller for the memory write request. The memory access intercept circuit stores the write data in a write data buffer, and, upon intercepting the proxy write data from the memory controller directed to a physical (PHY) interface circuit, retrieves the write data from the write data buffer and transmits the write data to the PHY interface circuit. The memory access intercept circuit subsequently receives, from the PHY interface circuit, loopback data, and stores the loopback data in a read data buffer.

Abstract: In a processing system, a conversion circuit coupled to a system bus generates a flow control unit (FLIT) and provides the FLIT to a link interface circuit for transmission over an external link. The external link may be a peripheral component interface (PCI) express (PCIe) link coupled to an external device comprising a cache or memory. The conversion circuit generates the FLIT, including write information based on the write instruction, metadata associated with at least one cache line, and cache line chunks, including bytes of a cache line. The cache line chunks may be chunks of one of the at least one cache line. Including the metadata in the FLIT avoids separately transmitting the at least one cache line and the metadata over the external link, which improves performance compared to generating separate transmissions. In some examples, the FLIT corresponds to a compute express link (CXL) protocol FLIT.

Abstract: Methods and systems for an advanced initialization bus (AIB) are presented. In an aspect, an AIB master sends, to an AIB slave, a serial clock over a first signal line, and performs a read operation with the AIB slave. Performing the read operation comprises sending a read command to the AIB slave via a bus comprising at least one bidirectional input/output (I/O) channel, each I/O channel having its own respective signal line, sending a read address to the AIB slave via the bus, receiving a copy of the serial clock from the AIB slave over a second signal line, and latching read data provided by the AIB slave via the bus into a read buffer using the copy of the serial clock as a data strobe. Thus, the AIB master latches the read data provided by the AIB slave using a read strobe also provided by the AIB slave.

Abstract: In exemplary aspects, to extend the measured boot process performed by a trusted platform module (TPM) circuit to earlier, primitive boot components that are processed before the TPM circuit becomes available to perform boot measurements, a secure boot processing system is configured to measure earlier, primitive boot components. The measured primitive boot components are used to update a virtual configuration register (CR) value in a final virtual CR. The TPM circuit uses the final virtual CR value as an initial starting CR value to measure subsequent boot components to provide end-to-end security for boot operations. In this manner, the final virtual CR value protects boot integrity of boot operations of its CPU even if they occur before availability of the TPM circuit.

Abstract: An integrated circuit (IC) employs error codes based on fields of data for protecting data transferred from a first circuit to a second circuit on the IC. Each bit of a generated error code is based on one or more fields of the data rather than on consecutive signal bits of a bus. Upon receiving the data in a second circuit, the error code is employed to determine whether the data has been transferred without an error. In case of an error, a response circuit generates an error signal having an error type corresponding to the data fields in which errors are detected. In some examples, the transferred data comprises a transaction request and the error signal indicates whether the transaction request has failed, the transaction request may be retried, or the transaction request may be completed despite the error.

Abstract: A compute node capable of enhanced performance and/or energy savings is proposed. The proposed compute node may check whether a last instruction of a first group—retrieved in a first decode cycle—is potentially a fusible instruction. If so, the proposed compute node may refrain from decoding the last instruction in the first decode cycle. Instead, the proposed compute node may determine if a first instruction of a second group of instructions retrieved in a second decode cycle (subsequent to the first decode cycle) is fusible with the last instruction of the first group. If so, the two instructions may be fused to a single micro-operation.

Abstract: Performing instruction fetch pipeline synchronization (IFPS) in processor-based devices is disclosed herein. In some exemplary aspects, a processor-based device provides multiple processors including a remote processor. The remote processor receives, from an issuing processor, a translation lookaside buffer (TLB) invalidation (TLBI) request indicating a request to invalidate an address translation, and subsequently receives an IFPS request from the issuing processor. The remote processor determines that any previously received TLBI requests including the most recent TLBI request have completed. Upon receiving the IFPS request, the remote processor determines that all instructions within a fetch pipeline portion that were potentially fetched using address translations older than the IFPS request have proceeded from the fetch pipeline portion of an instruction processing circuit to an execution pipeline portion of the instruction processing circuit.

Abstract: Mitigation of return stack buffer side channel attacks in a processor. Detecting a side channel attack or a fault in a return from a function call in the processor includes receiving a return exception level indication (or e.g., a return security level indication) indicating the exception level associated with the return and comparing the exception level associated with the return to the exception level (or security level) associated with the return address. The return exception level indicator may be received in conjunction with a return indication. The processing circuit accesses the first entry of the return stack buffer, which indicates the return address of the function call, and also accesses an exception level associated with the return address. The processing circuit compares the exception level associated with the return address to the exception level associated with the return to determine whether to use the return address in a prediction of instruction flow.

Abstract: A memory control circuit stores codewords in a memory module configured to limit errors to one wire of a memory interface. A codeword includes a first block with a first data portion and a second block with a second data portion. An error correction code symbol in the second block is used to locate and correct errors in the second block. Some bits of the first block are repurposed as metadata. The remaining bits are used as parity bits for detecting errors in the first block. The second block is merged with the first block before being stored to memory and demerged from the first block in a memory read operation. Demerging causes errors in the first block to create errors in a corresponding location in the second block. The location of errors found in the second block is used to locate and correct parity errors in the first block.

Abstract: A component die validation built-in self-test (VBIST) engine is presented. In an aspect, a component die includes component circuitry for performing a component function, interface circuitry for communicating with another die, and a VBIST circuit. The VBIST circuit includes a traffic generator that generates test data streams, a tracker that receives and validates test data streams, and a configurable switching matrix for coupling the traffic generator to at least one of the component circuitry, the interface circuitry, or the tracker, and for coupling at least one of the component circuitry, the interface circuitry, or the traffic generator to the tracker. The VBIST circuit can send traffic to and from the component circuitry directly, or indirectly via the interface circuitry in loopback mode, and can be used for memory initialization and test.

Abstract: A system and method are provided that enables a processor to undergo a functional test of its circuits prior to attachment to a semiconductor package and prior to use in a computing platform. The method of testing chips includes attaching a non-packaged semiconductor circuit to a test bed, loading computer instructions into a memory of the non-packaged semiconductor circuit, and operating the non-packaged semiconductor circuit in a test boot mode and executing the computer instructions in that mode. The operation logs one or more events of the execution of the instructions in the test boot mode and transmits the logs of the one or more events from the non-packaged semiconductor circuit via a joint test action group (JTAG) interface or a universal asynchronous receiver/transmitter (UART) interface.

Abstract: A system and method are described herein for estimating power usage of various components of a CPU and controlling voltage regulators based on the estimated power usage. The power estimates may be based on digital power meter readings at each component, on voltage information from a voltage regulator, and on other power information. This power information is transmitted over a mesh interconnect disposed throughout the CPU such that power estimation can be accurately calculated and used to control voltage regulators without being limited by external bus speeds. More of the power management processes and components may be disposed on the CPU and connected to the mesh interconnect. These power management processes include various calibrations, adjustments, and limits so as efficiently manage and use the more rapidly processed power estimations.

Abstract: A codeword read from memory includes data blocks including data and supplemental blocks including error correction code (ECC) symbols for detecting and correcting data errors. Metadata can be stored in the supplemental blocks to increase memory utilization but using bits of the supplemental blocks for metadata leaves too few bits remaining for the ECC symbols. To maintain error protection, the supplemental blocks include ECC symbols to protect a first data portion of the codeword and parity bits configured to protect a second data portion of the codeword. Errors in the first data portion can be located and corrected using the ECC symbols. Errors in the second data portion can be detected by the parity. For example, the first data portion is encoded based on the second data portion, so locations of parity errors correspond to locations of symbol errors, and parity errors can be corrected.

Abstract: Address range memory mirroring in a computer system, and related methods and computer-readable media. The computer system includes one or more memory mirror agents that are each configured to be programmed to mirror write data of a write request to a memory address mapped to the memory mirror agent. The memory mirror agent is configured to mirror write data to a redundant memory space in memory if the write memory address is within a programmed memory space to be mirrored by the memory mirror agent. The memory mirror agent can be programmed to perform memory mirroring based on specific address ranges to provide flexibility in controlling and changing the exact memory space of the memory system to be mirrored. If an error is detected in read data in response to a memory read request, the memory mirror agent can retrieve the stored redundant data to maintain data integrity.