Methods and apparatus for performing machine learning tasks, and in particular, to a neural-network-processing architecture and circuits for improved handling of partial accumulation results in weight-stationary operations, such as operations occurring in compute-in-memory (CIM) processing elements (PEs). One example PE circuit for machine learning generally includes an accumulator circuit, a flip-flop array having an input coupled to an output of the accumulator circuit, a write register, and a first multiplexer having a first input coupled to an output of the write register, having a second input coupled to an output of the flip-flop array, and having an output coupled to a first input of the first accumulator circuit.

Embodiments are directed to elliptic curve cryptography scalar multiplications in a generic field with heavy pipelining between field operations. A bit width is determined of operands in data to be processed by a modular hardware block. It is checked whether the bit width of the operands matches a fixed bit width of the modular hardware block. In response to there being a match, the modular hardware block processes the operands. In response to there being a mismatch, the operands are modified to be accommodated by the fixed bit width of the modular hardware block.

Embodiments of a processor are disclosed for performing arithmetic operations on a machine independent number format. The processor may include a floating point unit, and a number unit. The number format may include a sign/exponent block, a length block, and multiple mantissa digits. The number unit may be configured to perform an operation on two operands by converting the digit format of each mantissa digit of each operand, to perform the operation using the converted mantissa digits, and then to convert each mantissa digit of the result of the operation back into the original digit format.

The present disclosure relates to a computing device for processing a multi-bit width value, an integrated circuit board card, a method, and a computer readable storage medium. The computing device may be included in a combined processing apparatus, and the combined processing apparatus may further include a general interconnection interface, and an other processing device. The computing device interacts with the other processing device to jointly complete a computing operation specified by a user. The combined processing apparatus may further include a storage device connected to an apparatus and the other processing device and configured to store data of the apparatus and the other processing device. The solution of the present disclosure can split the multi-bit width value so that the processing capability of the processor is not influenced by the bit width.

A processing cluster of a processing cluster array comprises a plurality of registers to store input values of vector input operands, the input values of at least some of the vector input operands having different bit lengths than those of other input values of other vector input operands, and a compute unit to execute a dot-product instruction with the vector input operands to perform a number of parallel multiply operations and an accumulate operation per 32-bit lane based on a bit length of the smallest-sized input value of a first vector input operand relative to the 32-bit lane.

Described herein is a graphics processor including a processing resource including a multiplier configured to multiply input associated with the instruction at one of a first plurality of bit widths, an adder configured to add a product output from the multiplier with an accumulator value at one of a second plurality of bit widths, and circuitry to select a first bit width of the first plurality of bit widths for the multiplier and a second bit width of the second plurality of bit widths for the adder.

The present disclosure relates to a method of storing, by a load and store circuit or other processing means, a variable precision floating point value to a memory address of a memory, the method comprising: reducing the bit length of the variable precision floating point value to no more than a size limit, and storing the variable precision floating point value to one of a plurality of storage zones in the memory, each of the plurality of storage zones having a storage space equal to or greater than the size limit (MBB).

A floating-point number in a first format representation is received. Based on an identification of a floating-point format type of the floating-point number, different components of the first format representation are identified. The different components of the first format representation are placed in corresponding components of a second format representation of the floating-point number, wherein a total number of bits of the second format representation is larger than a total number of bits of the first format representation. At least one of the components of the second format representation is padded with one or more zero bits. The floating-point number in the second format representation is stored in a register. A multiplication using the second format representation of the floating-point number is performed.

Various embodiments are provided for performing hybrid precision floating point format computation via a simplified superset floating point unit in a computing system. One or more inputs, represented as a plurality of floating point number formats, may be converted into a superset floating point format prior to computation by one or more simplified superset floating point units (ssFPUs). A compute operation may be performed on the one or more inputs represented as the superset floating point format using the one or more ssFPUs.

A floating-point number in a first format representation is received. Based on an identification of a floating-point format type of the floating-point number, different components of the first format representation are identified. The different components of the first format representation are placed in corresponding components of a second format representation of the floating-point number, wherein a total number of bits of the second format representation is larger than a total number of bits of the first format representation. At least one of the components of the second format representation is padded with one or more zero bits. The floating-point number in the second format representation is stored in a register. A multiplication using the second format representation of the floating-point number is performed.