Abstract: In one example an apparatus comprises a first input node to receive a first plaintext input, a second input node to receive a second plaintext input, a third input node to receive a random mask and an advanced encryption standard (AES) circuitry configurable to operate in one of a first mode in which the random mask is added to the first plaintext input during one or more computations to convert the first plaintext input to a first ciphertext output, or a second mode in which the first plaintext input is converted to a first ciphertext output and the second plaintext input is converted to a second ciphertext output without using the random mask. Other examples may be described.

Abstract: In one example an apparatus comprises a first input node to receive a first plaintext input, a second input node to receive a random mask, an advanced encryption standard (AES) engine configurable to operate in one of a first mode in which the random mask is added to the first plaintext input during one or more computations performed by the AES engine, or second mode in which the random mask is not added to the first plaintext input during one or more computations performed by the AES engine. Other examples may be described.

Abstract: One embodiment provides a graphics processor comprising a memory controller and a graphics processing resource coupled with the memory controller. The graphics processing resource includes circuitry configured to execute an instruction to perform a matrix operation on first input including weight data and second input including input activation data, generate intermediate data based on a result of the matrix operation, quantize the intermediate data to a floating-point format determined based on a statistical distribution of first output data, and output, as second output data, quantized intermediate data in a determined floating-point format.

Abstract: Embodiments herein relate to optimizing the operation of multiple integrated circuits (ICs) operating in parallel. In one aspect, the ICs are arranged in a voltage-stacked configuration, and an operating frequency of each IC is controlled using a tunable replica circuit to stabilize its voltage drop. The tunable replica circuit mimics a critical path on the IC. In another aspect, an IC is divided into top and bottom portions which are in respective voltage domains on a substrate. The substrate include a deep n-well region for the higher voltage domain. In another aspect, a physically unclonable function (PUF) is used to generate identifiers for each IC among a multiple ICs on a board. Entropy sources of the PUF generate bits of the identifiers. Unstable entropy sources are identified and their bits are masked out.

Abstract: Methods and apparatus to parallelize data decompression are disclosed. An example method selecting initial starting positions in a compressed data bitstream; adjusting a first one of the initial starting positions to determine a first adjusted starting position by decoding the bitstream starting at a training position in the bitstream, the decoding including traversing the bitstream from the training position as though first data located at the training position is a valid token; outputting first decoded data generated by decoding a first segment of the bitstream starting from the first adjusted starting position; and merging the first decoded data with second decoded data generated by decoding a second segment of the bitstream, the decoding of the second segment starting from a second position in the bitstream and being performed in parallel with the decoding of the first segment, and the second segment preceding the first segment in the bitstream.

Abstract: One embodiment provides for a graphics processing unit to accelerate machine-learning operations, the graphics processing unit comprising a multiprocessor having a single instruction, multiple thread (SIMT) architecture, the multiprocessor to execute at least one single instruction; and a first compute unit included within the multiprocessor, the at least one single instruction to cause the first compute unit to perform a two-dimensional matrix multiply and accumulate operation, wherein to perform the two-dimensional matrix multiply and accumulate operation includes to compute an intermediate product of 16-bit operands and to compute a 32-bit sum based on the intermediate product.

Abstract: One embodiment provides a graphics processor comprising a memory controller and a graphics processing resource coupled with the memory controller. The graphics processing resource includes circuitry configured to execute an instruction to perform a matrix operation on first input including weight data and second input including input activation data, generate intermediate data based on a result of the matrix operation, quantize the intermediate data to a floating-point format determined based on a statistical distribution of first output data, and output, as second output data, quantized intermediate data in a determined floating-point format.

Abstract: Physically unclonable functions response in memory cells is improved by transistor sizing, transistor threshold voltage (VT) and body bias in the memory cell to improve the reproducibility of the memory cell and multiple Sense Amplifiers (SA) per column to further enhance physically unclonable function entropy. A physically unclonable function exploits a large number of read-sequence-order combinations available in a physically unclonable function memory array to generate an exponentially large challenge-response pair space, without incurring the area and energy costs of hosting and operating an exponentially large memory array.

Abstract: Methods and apparatus to parallelize data decompression are disclosed. An example method selecting initial starting positions in a compressed data bitstream; adjusting a first one of the initial starting positions to determine a first adjusted starting position by decoding the bitstream starting at a training position in the bitstream, the decoding including traversing the bitstream from the training position as though first data located at the training position is a valid token; outputting first decoded data generated by decoding a first segment of the bitstream starting from the first adjusted starting position; and merging the first decoded data with second decoded data generated by decoding a second segment of the bitstream, the decoding of the second segment starting from a second position in the bitstream and being performed in parallel with the decoding of the first segment, and the second segment preceding the first segment in the bitstream.

Abstract: A processing apparatus is provided comprising a multiprocessor having a multithreaded architecture. The multiprocessor can execute at least one single instruction to perform parallel mixed precision matrix operations. In one embodiment the apparatus includes a memory interface and an array of multiprocessors coupled to the memory interface. At least one multiprocessor in the array of multiprocessors is configured to execute a fused multiply-add instruction in parallel across multiple threads.

Abstract: Methods and apparatus to parallelize data decompression are disclosed. An example method selecting initial starting positions in a compressed data bitstream; adjusting a first one of the initial starting positions to determine a first adjusted starting position by decoding the bitstream starting at a training position in the bitstream, the decoding including traversing the bitstream from the training position as though first data located at the training position is a valid token; outputting first decoded data generated by decoding a first segment of the bitstream starting from the first adjusted starting position; and merging the first decoded data with second decoded data generated by decoding a second segment of the bitstream, the decoding of the second segment starting from a second position in the bitstream and being performed in parallel with the decoding of the first segment, and the second segment preceding the first segment in the bitstream.

Abstract: A processing apparatus is provided comprising a multiprocessor having a multithreaded architecture. The multiprocessor can execute at least one single instruction to perform parallel mixed precision matrix operations. In one embodiment the apparatus includes a memory interface and an array of multiprocessors coupled to the memory interface. At least one multiprocessor in the array of multiprocessors is configured to execute a fused multiply-add instruction in parallel across multiple threads.

Abstract: One embodiment provides for a graphics processing unit to accelerate machine-learning operations, the graphics processing unit comprising a multiprocessor having a single instruction, multiple thread (SIMT) architecture, the multiprocessor to execute at least one single instruction; and a first compute unit included within the multiprocessor, the at least one single instruction to cause the first compute unit to perform a two-dimensional matrix multiply and accumulate operation, wherein to perform the two-dimensional matrix multiply and accumulate operation includes to compute a 32-bit intermediate product of 16-bit operands and to compute a 32-bit sum based on the 32-bit intermediate product.

Abstract: In one embodiment, an apparatus comprises a decompression engine to determine a plurality of tokens used to encode a block of data; populate a lookup table with at least two of the tokens in order of increasing token length; disable a first portion of the lookup table and enable a second portion of the lookup table based on a value of a payload of the block of data; and search for a match between a token and the payload in the second portion of the lookup table.

Abstract: A processing apparatus is provided comprising a multiprocessor having a multithreaded architecture. The multiprocessor can execute at least one single instruction to perform parallel mixed precision matrix operations. In one embodiment the apparatus includes a memory interface and an array of multiprocessors coupled to the memory interface. At least one multiprocessor in the array of multiprocessors is configured to execute a fused multiply-add instruction in parallel across multiple threads.

Abstract: Methods and apparatus to parallelize data decompression are disclosed. An example method selecting initial starting positions in a compressed data bitstream; adjusting a first one of the initial starting positions to determine a first adjusted starting position by decoding the bitstream starting at a training position in the bitstream, the decoding including traversing the bitstream from the training position as though first data located at the training position is a valid token; outputting first decoded data generated by decoding a first segment of the bitstream starting from the first adjusted starting position; and merging the first decoded data with second decoded data generated by decoding a second segment of the bitstream, the decoding of the second segment starting from a second position in the bitstream and being performed in parallel with the decoding of the first segment, and the second segment preceding the first segment in the bitstream.

Abstract: A processing apparatus is provided comprising a multiprocessor having a multithreaded architecture. The multiprocessor can execute at least one single instruction to perform parallel mixed precision matrix operations. In one embodiment the apparatus includes a memory interface and an array of multiprocessors coupled to the memory interface. At least one multiprocessor in the array of multiprocessors is configured to execute a fused multiply-add instruction in parallel across multiple threads.

Abstract: One embodiment provides for a graphics processing unit to accelerate machine-learning operations, the graphics processing unit comprising a multiprocessor having a single instruction, multiple thread (SIMT) architecture, the multiprocessor to execute at least one single instruction; and a first compute unit included within the multiprocessor, the at least one single instruction to cause the first compute unit to perform a two-dimensional matrix multiply and accumulate operation, wherein to perform the two-dimensional matrix multiply and accumulate operation includes to compute a 32-bit intermediate product of 16-bit operands and to compute a 32-bit sum based on the 32-bit intermediate product.

Abstract: Technologies for memory and I/O efficient operations on homomorphically encrypted data are disclosed. In the illustrative embodiment, a cloud compute device is to perform operations on homomorphically encrypted data. In order to reduce memory storage space and network and I/O bandwidth, ciphertext blocks can be manipulated as data structures, allowing operands for operations on a compute engine to be created on the fly as the compute engine is performing other operations, using orders of magnitude less storage space and bandwidth.

Abstract: A processing system includes a processing core and a hardware accelerator communicatively coupled to the processing core. The hardware accelerator includes a random number generator to generate a byte order indicator. The hardware accelerator also includes a first switching module communicatively coupled to the random value indicator generator. The switching module receives an byte sequence in an encryption round of the cryptographic operation and feeds a portion of the input byte sequence to one of a first substitute box (S-box) module or a second S-box module in view of a byte order indicator value generated by the random number generator.