I wish llms had foreheads so I could tattoo "don't hard code config values" on it
vnprc
vnprc@trianglebitdevs.org
npub16vzj...mhgd
CTV+CSFS, Hashpool, Triangle BitDevs
GM
Hashpool test instance no longer stores miner IP addresses. 🫡
The test pool is back online! The front end looks exactly the same but the back end is significantly prettier now.
Thanks to all the testers contributing hashrate. 🫶
i feel the rust pain
this shitty vps has been compiling for 47 minutes and it's not close to done
Recovered video from @Sovereign Engineering Cohort 5.
vibing server side websites in rust is surprisingly fun
I think I'll drop 32 zeroes from the units. We need to keep the gigantic 'bits' units to compare ehash of different denominations (important because hashrate is growing exponentially). But it's easy to set 1 ehash token equal to lots of bits of work so that we can have human readable units.
The fundamental issue is that our computer architectures are optimized to represent numbers in 64 bits of space. SHA256 hashes are 256 bits in length. So we need to shrink the range of share values that our ehash number system can represent by a factor of 4.
The most important value to include in this range is the network difficulty target. This how hard it is to find a block. It currently takes ~76 leading zero bits in a mining share to find a block. Let's call this 76 difficulty. The luckiest block hash ever found had 94 difficulty.
If we start our unit scale at 32 difficulty and add 64 bits that takes us right up to 96 difficulty, the largest share we can represent in a 64 bit number. 96 difficulty is four times luckier than the luckiest share ever found.
The lower end of the range, 33 difficulty, is equal to 8.59 gigabits of work. This is the sum of 2^33 + 2^32 + 2^31 ... and so on. My laptop CPU can find 33 diff shares every minute or so. They would each be worth 1 ehash. My Bitaxe Gamma does ~1.2 TH/s and it can crank out 40 diff shares every second or two. With the new eHash scale each of those Bitaxe Gamma shares is worth 128 eHash. A bitcoin block takes ~8.8 trillion eHash to find.
With this scale we can nicely capture the range of useful values of mining shares. When the network difficulty substantially increases in the not too distant future we can slide our measurable difficulty window up to represent higher difficulty shares and drop the lowest values off of our scale.