Low precision computers can be efficient at finding possible answers to search problems. However, sometimes the task demands finding better answers than a single low precision search. A computer system augments low precision computing with a small amount of high precision computing, to improve search quality with little additional computing.

Implementations of the disclosure provide logarithm and anti-logarithm operations on a hardware processor based on linear piecewise approximation. An example processor includes a piece wise linear log approximation circuit that receives an input of a floating-point number comprising a sign, an exponent and a mantissa. The piece wise linear log approximation circuit approximates a fractional portion of a fixed point number using a linear approximation of the mantissa of the floating-point number. The piece wise linear log approximation circuit also derives an integer from the exponent.

A data processing apparatus is provided, to calculate an at least partial square root of a floating point number having an exponent and significand. Recurrence circuitry performs one or more iterations of an iterative square root operation, each of the one or more iterations receiving an input at least partial square root and an input remainder to produce the at least partial square root and a remainder of performing the iterative square root operation. The recurrence circuitry provides the at least partial square root and the remainder as the input at least partial square root and the input remainder for a subsequent iteration of the iterative square root operation. The recurrence circuitry includes initialisation circuitry to provide the at least partial square root and the remainder after at least an initial iteration of the one or more iterations. The initialisation produces the remainder by performing a selection of one of a plurality of predetermined values in dependence on whether the exponent is odd or even.

Low precision computers can be efficient at finding possible answers to search problems. However, sometimes the task demands finding better answers than a single low precision search. A computer system augments low precision computing with a small amount of high precision computing, to improve search quality with little additional computing.

Embodiments of the present disclosure are based on a recognition that some processors are configured with instructions to compute logarithms and exponents (i.e. some processors include log and exp circuits). Embodiments of the present disclosure are further based on an insight that the use of the existing log and exp circuits could be extended to compute certain other functions by using the existing log and exp circuits to transform from a Cartesian to a logarithmic domain and vice versa and performing the actual computations of the functions in the logarithmic domain, which may be computationally easier than performing the computations in the Cartesian domain.

A method for calculating an error signal that enables a diagnosis of the correctness of a product, determined by a first multiplier unit, of a first factor and a second factor, the error signal being determined by a difference formation unit as the difference of a sum logarithm and a product logarithm.

A processor or other device, such as a programmable and/or massively parallel processor or other device, includes processing elements designed to perform arithmetic operations (possibly but not necessarily including, for example, one or more of addition, multiplication, subtraction, and division) on numerical values of low precision but high dynamic range (LPHDR arithmetic). Such a processor or other device may, for example, be implemented on a single chip. Whether or not implemented on a single chip, the number of LPHDR arithmetic elements in the processor or other device in certain embodiments of the present invention significantly exceeds (e.g., by at least 20 more than three times) the number of arithmetic elements, if any, in the processor or other device which are designed to perform high dynamic range arithmetic of traditional precision (such as 32 bit or 64 bit floating point arithmetic).

A processor or other device, such as a programmable and/or massively parallel processor or other device, includes processing elements designed to perform arithmetic operations (possibly but not necessarily including, for example, one or more of addition, multiplication, subtraction, and division) on numerical values of low precision but high dynamic range (LPHDR arithmetic). Such a processor or other device may, for example, be implemented on a single chip. Whether or not implemented on a single chip, the number of LPHDR arithmetic elements in the processor or other device in certain embodiments of the present invention significantly exceeds (e.g., by at least 20 more than three times) the number of arithmetic elements, if any, in the processor or other device which are designed to perform high dynamic range arithmetic of traditional precision (such as 32 bit or 64 bit floating point arithmetic).

A processor or other device, such as a programmable and/or massively parallel processor or other device, includes processing elements designed to perform arithmetic operations (possibly but not necessarily including, for example, one or more of addition, multiplication, subtraction, and division) on numerical values of low precision but high dynamic range (LPHDR arithmetic). Such a processor or other device may, for example, be implemented on a single chip. Whether or not implemented on a single chip, the number of LPHDR arithmetic elements in the processor or other device in certain embodiments of the present invention significantly exceeds (e.g., by at least 20 more than three times) the number of arithmetic elements, if any, in the processor or other device which are designed to perform high dynamic range arithmetic of traditional precision (such as 32 bit or 64 bit floating point arithmetic).

A non-volatile memory device for performing compute in memory operations for a neural network uses a three dimensional NOR architecture in which vertical NOR strings are formed of multiple memory cells connected in parallel between a source line and a bit line. Weights of the neural network are encoded as threshold voltages of the memory cells and activations are encoded as word line voltages applied to the memory cells of the NOR strings. The memory cells are operated in the subthreshold region, where the word line voltages are below the threshold voltages. The NOR structure naturally sums the resultant subthreshold currents of the individual memory cells to generate the product of the activations and the weights of the neural network by concurrently applying input voltages to multiple memory cells of a NOR string.