Function residual_forward

experimental/kernels/ops.cpp:342–356

}

void residual_forward(Context& ctx, float* out, float* inp1, float* inp2, int N){
  unsigned long n = static_cast<unsigned long>(N);
  setLogLevel(kError);
  Tensor inp1_i = createTensor(ctx, Shape{n}, kf32, inp1);
  Tensor inp2_i = createTensor(ctx, Shape{n}, kf32, inp2);
  Tensor out_o = createTensor(ctx, Shape{n}, kf32);
  std::promise<void> promise;
  std::future<void> future = promise.get_future();
  Kernel op = createKernel(ctx, {kShaderResidual, 256, kf32},
                           Bindings{inp1_i, inp2_i, out_o},
                           /* nWorkgroups */ {cdiv(n, 256), 1, 1});
  dispatchKernel(ctx, op, promise);
  wait(ctx, future);
  toCPU(ctx, out_o, out, n * sizeof(float));
}

void residual_backward(Context& ctx, float* dinp1, float* dinp2, float* dout, int N){
  unsigned long n = static_cast<unsigned long>(N);