When running PiTool at 1.75 and SteamVR at 0.2 you are setting the overall supersampling factor to
1.75^2 * 0.2 = 0.6125
So technically it matches the same render target resolution when you set PiTool to 1.0 and SteamVR to 62%. Could you notice a visual quality difference between this and 1x1 setting?
Just to point out that SS factor in SteamVR is linear (as it factors the number of pixels) while in PiTool is quadratic (as it factors the dimensions of the image).
62% SS in SteamVR means that the render target is smaller by 0.62 when counting the pixels or by 0.78 when considering the dimensions.
What I am trying to say that if you feel you matched 1x1 by 1.75x0.2 you maybe as well match it with 1x0.62.
But the question is which uses higher GPU usage? Because if 1.75*0.2 leads to the same visual fidelity as 1x1, but uses less GPU usage, you can up the 1.75x0.2 to maybe 1.75x0.4
Technically, lower factors always mean lower GPU usage. So running PiTool x SteamVR at 1.75x0.2 equals to 0.62 of pixel count of running them at 1x1 (it does not mean the render time will 1/0.62 faster though as the rendering time includes more then texturing). My point is that both numbers combined define the final recommended render target resolution and the render time only depends on it and nothing else.
The game does not know if the resolution it gets from OpenVR was derived as: 1.75^2 x 0.2 or 1.0 x 0.62 or 1.5 x 0.27 all of them lead to the same render target resolution and if the game does not modify it internally further it will render in all three cases the exactly same image (at the same res).