In case anyone was interested if ETH does change algo going to ProgPOW I did some testing with cards I had on hand and probably what most gamers/miners use Mid-range GPUs. Even a low-end GPU I had on hand for fun! Here's what I noticed
RX580's Min of 8.5-9mh/s~ max of 10.4~ Huge increase in power, lowest I saw was 160w - Highest 220w!
GTX 1060 GB 5.8-6mh/s~ max of 7mh/s Lowest power was 85w highest was 140w
RX 460 3.1mh/s minor .01mh/s changes in overclocking mem/core or modding BIOs
RX 460 (unlocked shaders) 3.5mh/s again no changes when overclocking, netting .4mh/s increase with more shaders.
RX580's did not see any improvements when solely overclocking the memory, all the way up to 2300mhz memory OC. Only started to see improvements in hashrate when overclocking the core from 1200 up to 1400. However, power consumption was insane at 1400core 1150mv used 220w GPU+Riser!
GTX 1060 3GB seems hampered by its 192 bit-bus GTX 1060 6GB might net a slightly high hashrate because of the extra shaders but the "GTX 1060ti" with GDDR5X might actually do well. Going from 100% power to 75% power saw very minor differences in hashrate. Only saw a major drop when I hit around 60% power. 65% power + core overclock seemed to be the "sweet spot"
RX 460 only saw an increase with increasing the shaders and saw no improvement when OCing memory just like RX 580. Seems the 128bit-bus hampers greatly. PROGPOW requires high-bus such as HBM and many shaders. Core speed only counts when it's very low, shader count seems more important.
TLDR;
RX 580 increases in power + 50%
GTX1060 1.2mh/s difference in speed between 100% power and 60%
RX 460 unlocked shader helped increase hashrate. Overclocking the GPU did nothing.
Full results
https://docs.google.com/document/d/1GfMGWJOp_sf0WmrYNU8JWKgZXx6n1iQDBFgUmVi-4Co/edit?usp=sharingSide note
https://www.amd.com/en/technologies/blockchain