Function BM_Sum3DYCumsumGPU

tensorflow/core/kernels/scan_ops_test.cc:121–123

BENCHMARK(BM_Sum2DColumnCumsumGPU)->RangePair(1, 8192, 1, 8192);

static void BM_Sum3DYCumsumGPU(int iters, int num_x, int num_y) {
  Do3DYCumsum(iters, "gpu", num_x, num_y);
}
BENCHMARK(BM_Sum3DYCumsumGPU)->RangePair(64, 4096, 64, 4096);

static void BM_OneDCumsumGPU_reverse(int iters, int num_x) {