Hi,

I have 3 tensors, and I would like to compute this:

```
a = torch.randn(1, 32, 1, 1)
x = torch.randn(16, 32, 64, 64)
b = torch.randn(1, 32, 1, 1)
out = a *x + b
```

However, I found that this `a *x + b`

is not built-in function which makes inference not as fast as expected. So is there any built in function that can compute this in the c++ side like this ?

```
out = some_func(a, x, b)
```