The best setting was suggested by another user 2 weeks ago:
675/200 530W 803Mhs 0.66W/Mhs
760/250 790W 1008Mhs 0.783W/Mhs
FYI I have an EVGA 1300W gold.
Could you locate that post please? Your's seems to be the first reference to 0.66W/Mhs on this or the S3+ forums when I search 0.66W/Mhs.
Looking at the MHs i think it is C1. No S3+ can do 800 MHs. And it's to high for S5.
So i'm sticking with guess of C1
I ws looking at the .66 which would be watts per gigahash not megahash? The efficiency would seem very good unless you're off by the difference between megahash and gigahash which then would be quite bad. I looked on the S3+ for the frequency and voltage which might correspond as S3's have the same chips as the C1.
But I couldn't find your reference.
Thanks for warning about MHs. I did make a mistake on not using GHs.
And your correct about same chips. I would guess we will be close to S3 numbers, but cannot say for sure this is one miner I have not underclocked yet.
The numbers you seem to have faith in claim that at 67% of normal wattage he's getting only a 79% decrease in hashrate. That's even if his wattage is correct at normal frequency and voltage.
At normal frequency and voltage he claims 790watts. At normal frequency and voltage after an hour I had 1008GH/s but at 818 watts and after 2 hours I had 1002.88 @ 819watts.
I see from numbers on the S5 forum that my cheap power supplies run about 8% less efficient. If I reduce my power usage by 8% I get 753 watts. So, maybe my supply isn't 8% inefficient.
At 250M/0760 I got 1233GH/s/w or .811w/GH/s.
At 250M/0675 I got 1228GH/s/w or .814w/GH/s
At 243.75M/0675 I got 1242GH/s/w or .8059w/GH/s
So, you can see why I find it incredible that somebody went from .783 down to .66 because we're talking about miner efficiency change not supply efficiency change as supply efficiency shouldn't change that much in that wattage span.
As I said previously, I did see better figures at 200M on an S3 at stock voltage but it was during the coolest part of the day and ambient temperature seems to be a critical factor when measuring efficiency.
I have not tried 200M/0675 yet but I will eventually.
Oh, and my pumps and radiator fans are on a supply divorced from the C1 and not included in the calculations.