In an embodiment, a processor may include an execution engine to execute a plurality of instructions, a memory to store a tagged data structure comprising a plurality of entries, and an eviction circuit. The eviction circuit may be to: generate a pseudo-random number responsive to an eviction request for the tagged data structure; in response to a determination that the pseudo-random number is outside of a valid eviction range for the plurality of entries, generate an alternative identifier by rotating through the valid eviction range, the valid eviction range comprising a range of numbers that are valid to identify victim entries of the tagged data structure; and evict a victim entry from the tagged data structure, the victim entry associated with the alternative identifier. Other embodiments are described and claimed.

Some embodiments include apparatuses and electrical models associated with the apparatus. One of the apparatuses includes a power control unit to monitor a power state of the apparatus for entry into a standby mode. The apparatus can include a two-level memory (2LM) hardware accelerator to, responsive to a notification from the power control unit of entry into the standby mode, flush dynamic random access memory (DRAM) content from a first memory part to a second memory part. The apparatus can include processing circuitry to determine memory utilization and move memory from a first memory portion to a second memory portion responsive to memory utilization exceeding a threshold. Other methods systems and apparatuses are described.

Power consumption can be reduced by using a different memory chip for background processing. For example, a computing device can determine that a first application executing on a device complies with an application programming interface (API) for using a smaller dynamic access memory (DRAM) chip of the device that is smaller and consumes less power than a main DRAM chip of the device. The computing device can determine the device is inactive. The computing device can perform, by the API, background processing for the first application using the memory chip.

Embodiments of the invention are directed to systems and methods for utilizing a multi-tiered caching architecture in a multi-tenant caching system. A portion of the in-memory cache may be allocated as dedicated shares (e.g., dedicated allocations) that are each dedicated to a particular tenant, while another portion of the in-memory cache (e.g., a shared allocation) can be shared by all tenants in the system. When a threshold period of time has elapsed since data stored in a dedicated allocation has last been accessed, the data may be migrated to the shared allocation. If data is accessed from the shared allocation, it may be migrated back to the dedicated allocation Utilizing the techniques for providing a multi-tiered approach to a multi-tenant caching system can increase performance and decrease latency with respect to conventional caching systems.

An energy-efficient sequencer comprising inline multipliers and adders causes a read source that contains matching values to output an enable signal to enable a data item prior to using a multiplier to multiply the data item with a weight to obtain a product for use in a matrix-multiplication in hardware. A second enable signal causes the output to be written to the data item.

A peripheral proxy subsystem is placed between multiple hosts, each having a root controller, and single root I/O virtualization (SR-IOV) peripheral devices that are to be shared. The peripheral proxy subsystem provides a root controller for coupling to the endpoint of the SR-IOV peripheral device or devices and multiple endpoints for coupling to the root controllers of the hosts. The peripheral proxy subsystem maps the virtual functions of an SR-IOV peripheral device to the multiple endpoints as desired to allow the virtual functions to be allocated to the hosts. The physical function of the SR-IOV peripheral device is managed by the peripheral proxy device to provide the desired number of virtual functions. The virtual functions of the SR-IOV peripheral device are then presented to the appropriate host as a physical function or a virtual function.

A system and a method to allocate data to a first cache increments a first counter if a reuse indicator for the data indicates that the data is likely to be reused and decremented the counter if the reuse indicator for the data indicates that the data is likely not to be reused. A second counter is incremented upon eviction of the data from the second cache, which is a higher level cache than the first cache. The data is allocated to the first cache if the value of the first counter is equal to or greater than the first predetermined threshold or the value of the second counter equals zero, and the data is bypassed from the first cache if the value of the first counter is less than the first predetermined threshold and the value of the second counter is not equal to zero.

In an embodiment, a system may support programmable hashing of address bits at a plurality of levels of granularity to map memory addresses to memory controllers and ultimately at least to memory devices. The hashing may be programmed to distribute pages of memory across the memory controllers, and consecutive blocks of the page may be mapped to physically distant memory controllers. In an embodiment, address bits may be dropped from each level of granularity, forming a compacted pipe address to save power within the memory controller. In an embodiment, a memory folding scheme may be employed to reduce the number of active memory devices and/or memory controllers in the system when the full complement of memory is not needed.

Methods, systems, and devices for dynamic superblocks are described. In some examples, a superblock may be established across one or more dice of a memory device. A superblock may include one or more blocks from a plurality of planes of a memory die, and may be associated with a first performance cursor or a second performance cursor. The superblock may be established based on one or more criteria, such as a quantity of available blocks in a plane, a quantity of access operations performed on one or more blocks in a plane, or other criteria. Establishing a superblock associated with a first performance cursor may allow for performance criteria established by a host device to be maintained, while establishing a superblock associated with a second performance cursor may allow for garbage collection, wear leveling, and other maintenance operations to be performed on the memory device.

In various embodiments, a computing system includes, for example, a plurality of processing units that share access to a system cache. A cache management application receives, for example, resource savings information for each processing unit. The resource savings information indicates, for example, amounts of a resource (e.g., power) that are saved when different units of the system cache are allocated to a processing unit. The cache management application determines, for example, the number of units of system cache to allocate to each processing unit based on the received resource savings information.