I already did that, only the "GPU" the scene described above (Victoria 6 naked with hair, face close up) took 4 min. 2 seconds. 1920x1080 IRAY. With the "CPU" and "GPU" It took 4 minutes 12 seconds. 1920x1080 IRAY. I'll upload a video later.
The hair could actually be causing a problem. There is a known unfixed bug that causes issues in Iray when layers of geometry that use opacity maps intersect, as you'll find on some hair models. Most of my hair models seem to work fine though.
In general, I've found Octane and Iray to be close enough in rendering speed that I didn't notice a difference when I switched from Octane to Iray a couple months ago. (I never timed the renderers head to head, but if there is a speed difference between them on my Titan it isn't noticeable enough to affect my workflow.) The lack of crashing and better integration of Iray immediately began saving me time and lead me to make the switch, along with the presumption that all future DAZ products will come packaged with shaders ready to render in Iray which should save me a ton of time setting up materials. I've spent hours trying to get all of Stonemason's Urban Sprawl 2 materials set up properly for Octane, and there are just so many nodes in a set like that that it's a major headache and I never fully finished. (Simply organizing that many nodes in Octane is incredibly daunting before you even do anything to them.) Knowing that when Urban Sprawl 3 is released later this year it'll have Iray shaders already meticulously set up by Stonemason himself... well, it'll be beautiful.
Though I'm currently using Iray, I check in here just about daily hoping to see an update to Octane (or maybe some official news) which might give me a reason to fire up Octane again. I'll go with whichever software does the job best, so I'd love to see the Octane DS plugin be better than Iray, or at least a reliable stable companion to it when it is the better tool for a certain job.
Any future third party DAZ plugin renderer that would challenge the integration, speed, quality, and cost of Iray would have to be a totally next level of technology, akin to how much better Octane and Iray are than 3Delight and Reality. Being only marginally faster than Iray won't be enough to make up for the weaknesses in stability, support, and integration that comes with any 3rd party plugin renderer. A future Iray replacement would have to be another game-changer, doing something radically innovative that Iray can't be made to do, which I can't even currently imagine. Perhaps something which takes unique advantage of a class of computing hardware that doesn't yet exist.