You know how it was right to slow-grow Fish Hills for fastest Oracle? I think the same kind of idea applies to a Duckweed-Pyramids in CC with forge+stone. This is because it has only two good hammers tiles to work, and it can pass off half of its food resources to Fish Hills.
Suppose CC has been whipped down to size 3 with 19 food in the box (worst-case scenario). We work Gmine and two clams (other tiles worked by FH) at 7 food/turn excess. Need 23 to grow, so 4 turns with 5 food overflow. Meanwhile, 16 base hammers went on the 'mids.
At size 4 we work Pmine, Gmine and two clams (other tiles worked by FH) at 5 food/turn excess. Need 19 to grow, so 4 turns with 1 food overflow. Meanwhile, 32 base hammers on the 'mids.
At size 5 we work Pmine, Gmine, two clams and a coast again at 5 food/turn excess. Need 25 to grow, so 5 turns with no food overflow. Meanwhile, 40 base hammers on the 'mids.
At size 6 we put a turn of stuff on a settler (three clams, corn and two mines puts 11+8*1.25=21 hammers on it).
Next turn, we 3-whip it while working 2 clams and Gmine again (3*45*1.25+21+4*1.25 = 194) which gets us 45 overflow - IIRC this doesn't get the forge bonus, because it already has it, but it does get stone, so about 90.
So over those 15 turns we put 16+32+40=88 base hammers on, which is 198 after multipliers, to go with that 90 overflow. So 288 hammers in 15 turns. Three full cycles would build the Pyramids(750) - in practice two cycles and a final whip on about turn 40 (less with some overflow in at the start). Depending on the food and tile situation, some turns working a workshop can happen.
Note
- that a slow-build of the 'mids using two mines and two workshops would take 34 turns, so the Duckweed is comparable even before we factor in the settlers we built
- that Fish Hills had the use of corn and one clam throughout, and four turns of the Pmine per cycle
- a clear majority of the 'mids hammers came from working those mines
- we can't really get any more hammers from whip overflows
- that faster growth in CC is possible, but it doesn't have many good
tiles to work - so we do best to maximise use of the mines so long as we grow fast enough to 3-whip the settler after it has had a turn building
- my maths might be slightly out, but the principle holds
- this principle will also hold during the T105-T120 phase when we might build a forge - slow growth hastens the forge, so long as Fish Hills works the food tiles
- however, the need to give tiles to Fish Hills to work means it can't be running very many scientists (were we to follow that idea bcool had)
Still can't add attachments 
in FH (maybe) and CC. Growing into the expected happiness is a much stronger play than whipping the pyramids I think. Since we would immediately have excess pop to run as specialists.
