Why wouldn't the comparison results be invalid? Let's say a chart simply compare a 32G iPhone vs 32G Pixel NAND performance. That result would be very different if it was 128G iPhone vs 32G Pixel. Also that means we don't know how 32G Pixel is vs 128G Pixel is or how 32G vs 64G vs 128G S7 is. Sites doing these benchmarks don't compares variations usually. But if what you say is true that the typical difference in performance between sizes can be as large as 32G iPhone7 vs 128G (8-10x), all the charts out there comparing NAND speed in reviews are basically useless. The link you shown only have 2-3x difference for the worst case, most are <1 to 2x.
Fair point, but what sites aren't disclosing what sized memory device they're testing?
What I'm saying is true though, look at any SSD review where different sizes are tested.
http://techreport.com/review/29733/transcend-ssd370-solid-state-drive-reviewed/2
http://www.anandtech.com/show/10674/the-toshiba-ocz-vx500-ssd-review
https://www.pcper.com/reviews/Stora...-250GB-25-SATA-SSD-Full-Review-Return-2D-NAND
Bigger the drive the faster it'll be.
TR has a brief article too on the iPhone
Still, the difference seems to be a result of well-understood limitations in flash storage architectures, not because Apple "went a little bit chintzier in terms of the quality or caliber" of the 32GB phone's storage.