triton.experimental.gluon.language.amd.cdna5.wmma

triton.experimental.gluon.language.amd.cdna5.wmma(a, b, acc, _semantic=None)

Computes matrix multiplication a * b + acc using an AMD WMMA instruction.

Parameters:

• a (tensor) – The operand a to be multiplied.

• b (tensor) – The operand b to be multiplied.

• acc (tensor) – The accumulator tensor.