Isn't computational cost part of "all values"? From my understanding, we can't divide the "PoW Produced (24h)" with hashrate to find which one has the most effort, because of different hashing algorithms.
My point is that if you want to compute cost you should do it based on hashrate. Basically the method is to take the total hashrate and then come up with an estimation of the hardware that is used to produce that hashrate (GPUs and ASICs in case of ETH for example) then try to come up with an estimation of electricity cost and other possible costs and finally compute the total cost based on previous values.
ETH's PoW is not SHA3, it's ethash.
Ethash uses SHA3 or Keccak256 under the hood!
along with many other MUCH more expensive computations (computing the DAG whose values are sampled during an ethash hash involves tons of SHA3 itself, but that is done offline once every 30000 blocks).
Only updated every 125 hours!
It's embarrassing how you don't even know what order of magnitude means.
It is not embarrassing when English is my second language.
I have to benchmark the algorithm to know more, which I won't waste time on; but the code[1] only shows 2 SHA3 computations on random chunks with additional steps that don't look computationally expensive (only expensive memory-wise). However expensive it is, I don't get how you come up with 10k!
[1]