* @brief Return stride of descriptor on a single dimension * * @param dim Dimension to retrieve * @return Stride of dimension */

* Compute absolute value of every element in the tensor. For complex numbers * this returns the magnitude, or sqrt(x^2+y^2) * @param t * Tensor o

Utility functions

* Lazy assignment operator=. Used to create a "set" object for deferred * execution on a device * * @param op * Tensor view sour

fltflt_abs() returns the absolute value of a. This function assumes that a is normalized and thus that the sign of the hi component is the same as the

fltflt_fma() computes a * b + c with two normalizations. This is more efficient than fltflt_add(fltflt_mul(a, b), c), which uses three normalizations.

fltflt_round_toward_zero() truncates a fltflt value to an integer with the result being truncated toward zero (vs floor, which will truncate toward ne

* @brief Format timestamp in ISO 8601 format */

Getters

* @brief Convert log level to string */

* @brief Parse log level from string */

* Compute the t^t2 of two operators or tensors * @param t * LHS tensor or operator input * @param t2 * RHS tensor or operator power

Print a demangled stack backtrace of the caller function to FILE* out. */

A hard-coded JSON printer that generates a ctest resource-specification file: https://cmake.org/cmake/help/latest/manual/ctest.1.html#resource-specifi

Helper function to free all MatX caches. This function frees caches created for FFT plans, cuBLAS handles, and other state required for MatX transform

* @brief Determine if a pointer is printable by the device * * Pointers are printable if they're either a managed or device memory pointer * * @