Package io.github.kirstenali.deepj.tensor

Class CpuNdMatmul

java.lang.Object

io.github.kirstenali.deepj.tensor.CpuNdMatmul

public final class CpuNdMatmul
extends Object

General strided/broadcast CPU matmul; the existing FLOAT32 matrix fast path remains separate. Operands are staged once into contiguous FLOAT32 arrays so compact or strided inputs run the same parallel row kernel as dense FLOAT32; accumulation is FLOAT32 and only the result is rounded.

Method Summary

Modifier and Type	Method	Description
static Tensor	apply(Tensor left, Tensor right)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

apply

public static Tensor apply(Tensor left,
 Tensor right)