Canon were always very specific in stating dual pixel and dual gain are very different technologies.
I still find their marketing on this confusing.
The DGO whitepaper at least clears a few things up. Mainly the term photosite, which they use for the entire unit of two photodiodes. The marketing had made that sound like the individual photodiodes to me.
From the whitepaper I understand that each of the two photodiodes is read separately and they sequentially pass through the same amp, but set to different gains. It certainly is different to the DPAF in some sense, but not as much as some of the marketing would have you believe. After all, it relies on the two halfes pixels to work at all.
And it doesn't seem to actually amplify each photodiode with two different gains like some of the articles comparing it to the Arri implementation suggests.