So, in summary, you're generating a continuous ordered chain through a single CPU core before consensus, which is a serialized process passed between random leaders. Then you're feeding an arbitrary number of slices from said chain into parallel GPU cores to generate consensus through PoS. Is this right?
So the finality is probabilistic and is relatively slow? Which means that with complex smart contracts the actual throughput will be a magnitude lower than what’s being claimed? Which in its turn means that the only high throughput case for Solana is a simple currency transfers?
Apologies for the late reply Ivan. Smart Contracts go through the same pipeline as payments. The real question is what is deemed to be sufficient 'confirmations' on the network before it can be considered final. At the moment 1 confirmation is configured to around 800ms, we think this could be reduced to 400ms. So if we took BTC's definition of finality as an example (6 confirmations) it would take 6 x 800ms = 4.8seconds to reach 'finality. Ultimately it really depends on the application. I.e. is it mission critical, somewhere in between, or maybe even less. Hope that helps!
4:02 I have now read and watched several descriptions of this "Proof of History" idea. As far as I can tell, they are simply describing a block chain. Yes, duh, we can hash previous transaction hashes into the next transaction hash, proving an ordering. How is this anything new? This is literally how all blockchains work. Holding out hope that I am missing something here later in the video.
i think you are sorta correct. from my understanding so far, Proof of Stake is the primary consensus determiner, but that by introducing Proof of History, it does require the expenditure of some computational resources to be a validator, because as the network activity grows, you'll have to be able to keep up w/ the speed of the networks via some GPU racing. better than ETH because w/ ETH, there is NO cost to being a validator, which creates centralizing pressure. for SOL, it appears you do have to invest computational resources to be able to uphold your role as a validator
additionally, i believe the "Proof of History" creates a sort of objective "block height". the more proof of history/proof of work done, the more confident you can be that is the correct chain tip
AFAICT like Proof of History is basically a Proof of Work without adjustable difficulty, making Solana a hybrid PoW/PoS chain. This seems like it could work for now, but without adjustable difficulty, how can you trust that sha256 won't get 1000x faster or something? Wouldn't all the validator nodes need to constantly upgrade to the fastest hardware in order to make the timing function work? (I think I read that they just assume the nodes hash at a relatively similar rate, like within a ±30% range or something like that.) Seems like the same problem you get with Bitcoin & ASICs. Remember, these ASICs didn't even exist as an industry until the competitive pressures of Bitcoin brought them into being.
So, in summary, you're generating a continuous ordered chain through a single CPU core before consensus, which is a serialized process passed between random leaders. Then you're feeding an arbitrary number of slices from said chain into parallel GPU cores to generate consensus through PoS. Is this right?
Great summary. I would like to hear other's thoughts as well
So, recently they said the Solana clock is running 30 minutes behind. Can someone explain how this happened?
There was a block drift due to bugs and some validates using older versions of solana core,it's been fixed
So the finality is probabilistic and is relatively slow? Which means that with complex smart contracts the actual throughput will be a magnitude lower than what’s being claimed? Which in its turn means that the only high throughput case for Solana is a simple currency transfers?
Apologies for the late reply Ivan. Smart Contracts go through the same pipeline as payments. The real question is what is deemed to be sufficient 'confirmations' on the network before it can be considered final.
At the moment 1 confirmation is configured to around 800ms, we think this could be reduced to 400ms. So if we took BTC's definition of finality as an example (6 confirmations) it would take 6 x 800ms = 4.8seconds to reach 'finality.
Ultimately it really depends on the application. I.e. is it mission critical, somewhere in between, or maybe even less. Hope that helps!
Key takeaway: Alexander runs on a superior kind of operating system than to the average human being.
19:53
I can't understand about Tower BFT because My English is not good enough. so I leave timestamp to repeat for Tower BFT.
The coollest guys 🤩
4:02 I have now read and watched several descriptions of this "Proof of History" idea. As far as I can tell, they are simply describing a block chain. Yes, duh, we can hash previous transaction hashes into the next transaction hash, proving an ordering. How is this anything new? This is literally how all blockchains work. Holding out hope that I am missing something here later in the video.
i think you are sorta correct.
from my understanding so far, Proof of Stake is the primary consensus determiner, but that by introducing Proof of History, it does require the expenditure of some computational resources to be a validator, because as the network activity grows, you'll have to be able to keep up w/ the speed of the networks via some GPU racing.
better than ETH because w/ ETH, there is NO cost to being a validator, which creates centralizing pressure.
for SOL, it appears you do have to invest computational resources to be able to uphold your role as a validator
additionally, i believe the "Proof of History" creates a sort of objective "block height". the more proof of history/proof of work done, the more confident you can be that is the correct chain tip
AFAICT like Proof of History is basically a Proof of Work without adjustable difficulty, making Solana a hybrid PoW/PoS chain. This seems like it could work for now, but without adjustable difficulty, how can you trust that sha256 won't get 1000x faster or something? Wouldn't all the validator nodes need to constantly upgrade to the fastest hardware in order to make the timing function work? (I think I read that they just assume the nodes hash at a relatively similar rate, like within a ±30% range or something like that.) Seems like the same problem you get with Bitcoin & ASICs. Remember, these ASICs didn't even exist as an industry until the competitive pressures of Bitcoin brought them into being.
@@JamesBlacklock yeah their answer here is kinda weak and basically they just assume it won't be overrun lol
Yeah. Poh is basically pow.
near 🎉🎉🎉🎉🎉
Good to see Solana has a history. I was worried
Worried about what
@@kopenhavnkopenhavn5817He is worrying about Solana's proof of history.
does Solana chain actually exist?
no bro it's imaginary
Шкода що не побачив це відео в 2019
Awesome!! $SOL now is above $1.7 on #MXC_Exchange and it keeps going up! I think any time soon it will hit another all time high maybe $2! 🙌🤑
lol $2 dollars
@@victorcheg4556 glad i got it below $2. Still hodling
@@chiquioj3245 Well done! what's your prediction? when are you gonna sell SOL?
@@victorcheg4556 well i think it is better than eth (tech wise) id love to see it around 1k then ill decide
legend
Solana- The fastest blockchain ever invented
not really
Kadera and the others
I love the fact that transaction are really quick and transaction fee are extremely low
What about a bounty to catch anomalies? That would encourage some to go back in time with their GPU cycles.
бля, русские ребята, а по русски не говорят, вот как мне понять вас теперь?
Why erase the board, when the other 90% is empty?
Only 20% of the board is actually usable i.e. in easy reach for the guy to write on. Unless you want to start bringing in step ladders ;)
Hm... one of the worst usages of a whiteboard and general presentation, intonation, posture, structure ... ive seen in a while.
hard to understand for a newcomer
A shallow deep dive? Seriously?
lol