Abstract: Methods and systems for dynamic key-value cache management are provided. In one example, a method of processing key-value blocks (KV blocks) in a processor system includes providing a cyclic distribution of elements for a plurality of KV blocks associated with a key-value cache, the cyclic distribution providing a subset of elements of each of the plurality of KV blocks in one or more registers. In some implementations, the example method of processing key-value blocks (KV blocks) in a processor system includes accumulating a plurality of partial multiplication results to provide an output in a lossless numerical representation based at least in part on the cyclic distribution of the plurality of KV blocks.

Abstract: Circuitry for decomposing block floating-point numbers into lower precision floating-point numbers is provided. The circuitry may include a high precision storage circuit configured to provide high precision floating-point numbers, input selectors configured to receive the high precision floating-point numbers from the high precision storage circuit and to generate corresponding lower precision floating-point components with adjusted exponents, and a low precision block floating-point vector circuit configured to combine the various lower precision floating-point components generated by the input selectors. The lower precision floating-point components may be processed spatially or over multiple iterations over time.

Abstract: Circuitry for decomposing block floating-point numbers into lower precision floating-point numbers is provided. The circuitry may include a high precision storage circuit configured to provide high precision floating-point numbers, input selectors configured to receive the high precision floating-point numbers from the high precision storage circuit and to generate corresponding lower precision floating-point components with adjusted exponents, and a low precision block floating-point vector circuit configured to combine the various lower precision floating-point components generated by the input selectors. The lower precision floating-point components may be processed spatially or over multiple iterations over time.

Abstract: Circuitry for decomposing block floating-point numbers into lower precision floating-point numbers is provided. The circuitry may include a high precision storage circuit configured to provide high precision floating-point numbers, input selectors configured to receive the high precision floating-point numbers from the high precision storage circuit and to generate corresponding lower precision floating-point components with adjusted exponents, and a low precision block floating-point vector circuit configured to combine the various lower precision floating-point components generated by the input selectors. The lower precision floating-point components may be processed spatially or over multiple iterations over time.

Abstract: Circuitry for decomposing block floating-point numbers into lower precision floating-point numbers is provided. The circuitry may include a high precision storage circuit configured to provide high precision floating-point numbers, input selectors configured to receive the high precision floating-point numbers from the high precision storage circuit and to generate corresponding lower precision floating-point components with adjusted exponents, and a low precision block floating-point vector circuit configured to combine the various lower precision floating-point components generated by the input selectors. The lower precision floating-point components may be processed spatially or over multiple iterations over time.

Abstract: Circuitry for decomposing block floating-point numbers into lower precision floating-point numbers is provided. The circuitry may include a high precision storage circuit configured to provide high precision floating-point numbers, input selectors configured to receive the high precision floating-point numbers from the high precision storage circuit and to generate corresponding lower precision floating-point components with adjusted exponents, and a low precision block floating-point vector circuit configured to combine the various lower precision floating-point components generated by the input selectors. The lower precision floating-point components may be processed spatially or over multiple iterations over time.