Technologies for dynamically allocating resources within a self-managed node include a self-managed node to receive quality of service objective data indicative of a performance objective of one or more workloads assigned to the self-managed node. Each workload includes one or more tasks. The self-managed node is also to execute the one or more tasks to perform the one or more workloads, obtain telemetry data as the workloads are performed, determine, as a function of the telemetry data, an adjustment to the allocation of resources among the workloads to satisfy the performance objective, and apply the determined adjustment as the workloads are performed by the self-managed node. Other embodiments are also described and claimed.

Embodiments are generally directed apparatuses, methods, techniques and so forth to receive a sled manifest comprising identifiers for physical resources of a sled, receive results of an authentication and validation operations performed to authenticate and validate the physical resources of the sled, determine whether the results of the authentication and validation operations indicate the physical resources are authenticate or not authenticate. Further and in response to the determination that the results indicate the physical resources are authenticated, permit the physical resources to process a workload, and in response to the determination that the results indicate the physical resources are not authenticated, prevent the physical resources from processing the workload.

Devices and techniques are generally described for compression of natural language processing models. A first index value to a first address of a weight table may be stored in a hash table. The first address may store a first weight associated with a first feature of a natural language processing model. A second index value to a second address of the weight table may be stored in the hash table. The second address may store a second weight associated with a second feature of the natural language processing model. A first code associated with the first feature and comprising a first number of bits may be generated. A second code may be generated associated with the second feature and comprising a second number of bits greater than the first number of bits based on a magnitude of the second weight being greater than a magnitude of the first weight.

Technologies for a low-latency interface with data storage of a storage sled in a data center are disclosed. In the illustrative embodiment, a storage sled stores metadata including the location of data in a storage device in low-latency non-volatile memory. When accessing data, the storage sled may access the metadata on the low-latency non-volatile memory and then, based on the location determined by the access to the metadata, access the location of the data in the storage device. Such an approach results in only one access to the data storage in order to read the data instead of two.

Examples may include racks for a data center and sleds for the racks, the sleds arranged to house physical resources for the data center. The sleds and racks can be arranged to be autonomously manipulated, such as, by a robot. The sleds and racks can include features to facilitate automated installation, removal, maintenance, and manipulation by a robot.

An apparatus includes a first encoder circuit configured to compress a block of data using dictionary based compression and a second encoder circuit connected to the first encoder circuit to receive the compressed block of data from the first encoder circuit. The second encoder circuit is configured to further compress the compressed block of data according to a codebook. The codebook is based on a distribution of data of a prior block of data or a distribution of data of a portion of the block of data that is less than the block of data. The operation of the second encoder circuit overlaps with the operation of the first encoder circuit to achieve high throughput and avoid the need for a large block of memory (e.g., SRAM) to occupy the data in flight until the second encoder circuit can start.

A system and method for filesystem data compression using codebooks, that measures in real-time the probability distribution of an encoded data stream, compares the probability distribution to a reference probability distribution, and uses one or more statistical algorithms to determine the divergence between the two sets of probability distributions to determine if an unusual distribution is the result of a data intrusion. The system comprises both encoding and decoding machines, an intrusion detection module, a codebook training module, and various databases which perform various analyses on encoded data streams. Further, the system comprises a system for integrating the compression into a filesystem for both system-wide compression on a per-file or filegroup basis, and intrusion or alteration detection of files.

The generation of symbol-encoded data from digital data, as part of the compression of the digital data into a compressed digital data, can be performed with reference to multiple alternative alphabets. A selection of a specific alphabet is made based on the digital data being compressed, the compression parameters, or combinations thereof. Information indicative of the selected alphabet is encoded into one or more headers of the resulting compressed digital data. A single alphabet can be selected for all of a set of digital data being compressed, or multiple different alphabets can be selected, with different ones of the multiple different alphabets being utilized to compress different portions of the digital data. Additionally, rather than explicitly specifying a specific selected alphabet, the header information can comprise information from which a same alphabet can be independently selected heuristically by both the compressor and the corresponding decompressor.

Embodiments are generally directed apparatuses, methods, techniques and so forth to receive a sled manifest comprising identifiers for physical resources of a sled, receive results of an authentication and validation operations performed to authenticate and validate the physical resources of the sled, determine whether the results of the authentication and validation operations indicate the physical resources are authenticate or not authenticate. Further and in response to the determination that the results indicate the physical resources are authenticated, permit the physical resources to process a workload, and in response to the determination that the results indicate the physical resources are not authenticated, prevent the physical resources from processing the workload.

Technologies for dynamically managing resources in disaggregated accelerators include an accelerator. The accelerator includes acceleration circuitry with multiple logic portions, each capable of executing a different workload. Additionally, the accelerator includes communication circuitry to receive a workload to be executed by a logic portion of the accelerator and a dynamic resource allocation logic unit to identify a resource utilization threshold associated with one or more shared resources of the accelerator to be used by a logic portion in the execution of the workload, limit, as a function of the resource utilization threshold, the utilization of the one or more shared resources by the logic portion as the logic portion executes the workload, and subsequently adjust the resource utilization threshold as the workload is executed. Other embodiments are also described and claimed.