no worries, I just thought it was being leveraged in a way I didn't know.You are correct here. I was conflating DirectML with WinML.
Also, playing a bit of Devil's Advocate. I am actually leaning towards there being dedicated hardware for each of these in the next MS console, though I don't consider it certain.
I've never cared about compute power until I actually started training on my laptop. Real analysis is mega slow and I can see now why the pricing goes up like crazy. We have a PC with a GPU locked away and it's spending (I believe a Titan) a solid week grinding out its latest model for customer churn.
I've got a significantly smaller project on our team and we don't even have real gear.
This is a nightmare, my laptop won't stop crashing for such a small data set.
I don't have a GPU either lol those intel iGPUs aren't looking so silly to me anymore. Sigh.
Fortunately there is an available hadoop server in our organization but not yet sure what the limitations of it are.