147     std::array<size_t, XNN_MAX_TENSOR_DIMS> output_dims;  in TestF32()  local
150     std::fill(output_dims.begin(), output_dims.end(), 1);  in TestF32()
157       output_dims[i] = std::max(input1_dims[i], input2_dims[i]);  in TestF32()
160       std::accumulate(output_dims.begin(), output_dims.end(), size_t(1), std::multiplies<size_t>());  in TestF32()
173       output_stride *= output_dims[i - 1];  in TestF32()
186       for (size_t i = 0; i < output_dims[0]; i++) {  in TestF32()
187         for (size_t j = 0; j < output_dims[1]; j++) {  in TestF32()
188           for (size_t k = 0; k < output_dims[2]; k++) {  in TestF32()
189             for (size_t l = 0; l < output_dims[3]; l++) {  in TestF32()
190               for (size_t m = 0; m < output_dims[4]; m++) {  in TestF32()
191                 for (size_t n = 0; n < output_dims[5]; n++) {  in TestF32()
334       for (size_t i = 0; i < output_dims[0]; i++) {  in TestF32()
335         for (size_t j = 0; j < output_dims[1]; j++) {  in TestF32()
336           for (size_t k = 0; k < output_dims[2]; k++) {  in TestF32()
337             for (size_t l = 0; l < output_dims[3]; l++) {  in TestF32()
338               for (size_t m = 0; m < output_dims[4]; m++) {  in TestF32()
339                 for (size_t n = 0; n < output_dims[5]; n++) {  in TestF32()