The relation (hash)/(point addition) for devices

[BitcoinADAB]

How much is 1 Yottakeys/s for (Pollard / Rho) ?

For Pollard / Rho you need simple point additions.
Wikipedia: yotta 10^24 = 1000000000000000000000000
1 Yottakeys/s = 10^24 keys/s

To compare it with the Bitcoin network ~100 Ehash/s = 100000000000000000000 hash/s

1 Yottakeys/s = 10000 x more keys than hashes. But hashes are more complicated to calculate.

What would the relation (hash)/(point addition) be for devices?
Or, how many point additions could (if built for this purpose) calculate a miner, that let's say hashes with 1 Thashes/s = 1000000000000 hashes/s?

Does somebody know it?

[1714535827]

1714535827

[1714535827]

1714535827

[1714535827]

1714535827

[NotATether]

I don't think this question makes much sense because a hash is just a double SHA256 which is unrelated to EC point addition.

The closest relation would be between hashes and "keys" which is just a way of saying a bunch of point additions/multiplications and then a double SHA256 + RIPEMD160.

[BitcoinADAB]

I don't think this question makes much sense because a hash is just a double SHA256 which is unrelated to EC point addition.

The closest relation would be between hashes and "keys" which is just a way of saying a bunch of point additions/multiplications and then a double SHA256 + RIPEMD160.

Yes, they are unrelated. The question is, if an ASIC miner, that is a special device to mine, would be built for EC point addition, with approximately the same number of chips (of course, they would have a different architecture) etc., how many point additions would it make. In the example we have a miner that hashes with 1 Thashes/s = 1000000000000 hashes/s. What would this number be for an 'EC point addition ASIC'?

[NotATether]

Yes, they are unrelated. The question is, if an ASIC miner, that is a special device to mine, would be built for EC point addition, with approximately the same number of chips (of course, they would have a different architecture) etc., how much point additions would it make.

You can't estimate this value, unless you know how many logic operations are used in each chip and how the gates are connected to each other.

[BitcoinADAB]

...

You can't estimate this value, unless you know how many logic operations are used in each chip and how the gates are connected to each other.

Thanks. What we want is to calculate the energy consumption for 2^128 keys. We need this for the BitcoinADAB project https://bitcointalk.org/index.php?topic=5347791.0, that you know.
Miners have this efficiency Gh/j, so it would be interesting to have an approximate number for the EC point addition case.

[NotATether]

Thanks. What we want is to calculate the energy consumption for 2^128 keys. We need this for the BitcoinADAB project https://bitcointalk.org/index.php?topic=5347791.0, that you know.
Miners have this efficiency Gh/j, so it would be interesting to have an approximate number for the EC point addition case.

You should be looking to calculate Megahash/Watt (MegaKey/Watt in this case) since it's a more common metric for GPUs. In this case you must first design a program that benchmarks EC point operation speed on GPUs and prints their MKey value (you can fork Kangaroo and use the code in Check.cpp for this).

Then you can take 2^128 keys and divide it by the MKey/s for each GPU model tested to get the number of seconds (and from there the number of GPUs/total watts needed to search 2^128 keys in 1 second) a search of 2^128 keys will take.  

[BitcoinADAB]

...

Yes, that way it will work.

Then we can say, even without having the devices, if the project is possible or not.

[BitcoinADAB]

Then we can say, even without having the devices, if the project is possible or not.

Assumption:

We have EC point addition ASICs that have the same rate like for mining (1 hash = 1 key).
Miner power consumption: 30 Ghash/Wattsec
'EC point addition ASIC' power consumption: 30 Gkeys/Wattsec

We have to calculate 2^128 keys

World power consumption: 28,000 TWatthours     https://www.iea.org/reports/world-energy-outlook-2019/electricity

Result:

We would need to calculate ~13,000 years

nearly impossible, but not impossible

if we have a rate of 1,000 keys for 1 hash with the same power consumption: 13 years
and if we have a rate of 1,000,000 keys for 1 hash: ~3 years with the power consumption of the bitcoin network (120 TWatthours)

then it would become possible

We need improved 'EC point addition ASICs'

and remember ...

The hashrate of Bitcoin was in the first year ~10 Mhash/sec and 10 years later 100 Ehash/s. If someone told us in 2009 that in 2019 the network will hash 10,000,000,000,000 times faster, what would we say? 10,000,000,000,000 more participants? No, the hardware and hardware use improved.

Edit:
Bitcoin mining energy efficiency over time - we calculate EC point additions with CPU/GPU - we need ASICs

Bitcoin mining energy efficiency over time. The y-axis represents bitcoin mining energy efficiency in units of Watts per GH/s on a logarithmic scale. The x-axis represents time, beginning with the introduction of Bitcoin in 2009. Mining technology regimes are indicated as well.  

Edit:
We have received a DM that "The EC operations are several orders of magnitude more expensive than hashing."

[NotATether]

We need improved 'EC point addition ASICs'

Again, like I said, you need to have experience with designing logic gates to make an ECC ASIC, otherwise you'll just keep wandering in theory. Does anyone on your team possess this knowledge?

[BitcoinADAB]

We need improved 'EC point addition ASICs'

Again, like I said, you need to have experience with designing logic gates to make an ECC ASIC, otherwise you'll just keep wandering in theory. Does anyone on your team possess this knowledge?

We can't build ASICs. Our idea is, that ASIC producers will build it, if there is demand. Once the project starts, people will check it, if it is possible and if they can make a profit on it. And then there will be a large competition to have the best keys/Watt rate, so that we will have far better EC point addition rates than with today's GPUs.
But before starting such a project, we have to check the possibility. That energy consumption calculation helped us to say that it is possible for 2^128. We know that for 2^256 it is not possible, there is not enough energy to calculate it.