Function ggml_cl_can_mul_mat

ggml-opencl.cpp:1806–1821

bool ggml_cl_can_mul_mat(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * dst) {
    const int64_t ne10 = src1->ne[0];

    const int64_t ne0 = dst->ne[0];
    const int64_t ne1 = dst->ne[1];

    // TODO: find the optimal values for these
    if ((src0->type == GGML_TYPE_F32 || src0->type == GGML_TYPE_F16 || ggml_is_quantized(src0->type)) &&
        src1->type == GGML_TYPE_F32 &&
        dst->type == GGML_TYPE_F32 &&
        ((ne0 >= 32 && ne1 >= 32 && ne10 >= 32) || src0->backend == GGML_BACKEND_GPU)) {
        return true;
    }

    return false;
}

static bool ggml_cl_mul_mat_use_f16(const struct ggml_tensor * src0, const struct ggml_tensor * src1, struct ggml_tensor * /* dst */) {
    // If device doesn't support FP16