Let's see. One could model an RGBG display as an RGB display where a pair of pixels is missing one R and B pixel each. To compensate for this, the singular R and B pixels would need to be driven in a way as to double their maximum output. Alternatively, the G subpixels could be halved in surface area and, thus, luminosity.

Since chromatic acuity in one's eye is worse than luminous acuity, one would not lose image quality.

So, given a specific subpixel budget, a PenTile display would be better if the pixel density is high enough to test the limits of the human eye.

Did I miss anything yet?