Priya Goyal

@zdevito I think it will be useful to have this feature available. We stumbled upon one use case today when using ATen outside pytorch

Hi @ReyhaneAskari , 1. GPU used: M40 with 11443 MiB 2. Max image is computed by increasing the batch size until the model runs OOM. It will be N for...

yep, thanks for catching, updating the description :)

for more information, copy pasting some information from our internal discussion on this: Andrew: "The issue is that it requires us to do compute outside of the scope where we...

Hi @arogozhnikov , thanks for reporting. We don't support CPU computations in TC right now. However, adding a simple check for `is_cuda()` should definitely be possible. I'll re-title this issue...

I believe @nicolasvasilache would have something to add ;)

cc @jekbradbury

the reason is because we use Halide IR and it makes assumption that the tensor dependencies are strictly DAG. Hence it doesn't work in case of softmax variants that failed.

thanks @seongwook-ham for the report. Can you provide us a repro? It also depends on what operation you are running and sometimes compilation takes a lot of time that makes...

Hi @arogozhnikov , thanks for bringing this to attention. Indeed we should modify the documentation and display what works. Assigning this to myself to reflect these changes in docs :)...