Abstract: One embodiment provides an apparatus comprising a memory stack including multiple memory dies and a parallel processor including a plurality of multiprocessors. Each multiprocessor has a single instruction, multiple thread (SIMT) architecture, the parallel processor coupled to the memory stack via one or more memory interfaces. At least one multiprocessor comprises a multiply-accumulate circuit to perform multiply-accumulate operations on matrix data in a stage of a neural network implementation to produce a result matrix comprising a plurality of matrix data elements at a first precision, precision tracking logic to evaluate metrics associated with the matrix data elements and indicate if an optimization is to be performed for representing data at a second stage of the neural network implementation, and a numerical transform unit to dynamically perform a numerical transform operation on the matrix data elements based on the indication to produce transformed matrix data elements at a second precision.

Abstract: A header machine for drilling specified hole patterns in pipe and for cutting the pipe to a specified length to form headers for use in heat transfer products. The header machine includes hold clamps for selectively holding the pipe during the processing of the pipe and feed clamps for selectively holding the pipe during the feeding thereof through the header machine. The feed clamps are mounted on a feed carriage which is movably mounted for movement forwardly and rearwardly. The feed carriage is driven by a stepper motor operably coupled to the feed carriage for moving the feed carriage and the pipe endwise to various drilling positions. The header machine also includes a spud drill for drilling spud holes in the pipe, a connection drill for drilling connection holes in the pipe, a circular saw for cutting the pipe to a specified length, and provisions for making double circuit patterns of the spud hole.