Open vatai opened 2 years ago
this really should be the other way around backend first, layout came alter I'd assume at 1.4. there were none, we can surely do some try catch there
Thanks, I really wanted your input on this. It felt it should be the other way around... can it be that in old torch, the way to set mkldnn was .to_mkldnn
instead of setting backend
? But maybe I overlooked something
The two commits Jens compared:
e20f59d9c2326367eb7ce329b5a251e2ce61d1eb
works
5e67fe4a7111d2b40a6b9314290ca1ba809543e7
broken
Ok. It is something funky indeed. Both commits have backend.mkldnn.enable = True
etc. but for some reason the version that worked for Jens was not calling the backend.mkldnn.enable = True
code, despite having --backend=DNNL
set :cry:
why torch 1.4 btw? it's more ancient than dinosaurs
Yepp!!! :grin:
Jens reported this, so didn't check myself what's exactly happening. I suspect the following:
.to_mkldnn()
method), while newer version can also doobackend=mkldnn
separately...backend=mkldnn
when you just want the.to_mkldnn()
(i.e.--tensor_layout
).Some error messages: