BitShares Forum
Main => General Discussion => Topic started by: coolspeed on June 05, 2014, 11:02:56 am
-
Can the TPS of DPOS be as high as NXT Transparent Forging?
TPS = Transactions Per Second
Neither do I know whether NXT already implemented TF.
-
Can the TPS of DPOS be as high as NXT Transparent Forging?
TPS = Transactions Per Second
Neither do I know whether NXT already implemented TF.
TF is not implemented completely yet. If DPOS gives a way to predict next miner with high probability then the answer is YES, I believe.
-
in DPOS there are no Miners but Delegates .. the next delegate is fixed for one complete round .. after a round the order of delegates is randomly drawn again.. over and over .. round for round
I'd say, depending on the connection/network of the delegates DPOS can be really fast .. in the testnet block are faster than 30 secs
-
I'd say, depending on the connection/network of the delegates DPOS can be really fast .. in the testnet block are faster than 30 secs
TPS is more about throughput than latency.
-
Dan has said that dpos can scale to the size of visas 10,000 tps. I think nxt can handle 1,000 tps with tf, so yes dpos is better.
-
Dan has said that dpos can scale to the size of visas 10,000 tps. I think nxt can handle 1,000 tps with tf, so yes dpos is better.
Still unproven though
-
Obviously. But when you think about the fact that bitcoin network constitutes largest super computer in the world, the transition from using that level of computational resources for pointless work to using it for increased network capacity I'd say that dpos can scale well beyond visa in the future.
-
Obviously. But when you think about the fact that bitcoin network constitutes largest super computer in the world, the transition from using that level of computational resources for pointless work to using it for increased network capacity I'd say that dpos can scale well beyond visa in the future.
+5% (obviously ;D )
-
Can the TPS of DPOS be as high as NXT Transparent Forging?
TPS = Transactions Per Second
Neither do I know whether NXT already implemented TF.
TF is not implemented completely yet. If DPOS gives a way to predict next miner with high probability then the answer is YES, I believe.
The exact order for delegates is known between every shuffle. So except for between shuffles you always know the next delegate with certainty at least 30s in advance.
-
This is an issue of trust once you get to 10K/sec because the bandwidth required is no longer able to be supported by a decentralized network. However, DPOS allows for centralization of delegates without losing the property of decentralized control.
-
Can the TPS of DPOS be as high as NXT Transparent Forging?
TPS = Transactions Per Second
Neither do I know whether NXT already implemented TF.
TF is not implemented completely yet. If DPOS gives a way to predict next miner with high probability then the answer is YES, I believe.
The exact order for delegates is known between every shuffle. So except for between shuffles you always know the next delegate with certainty at least 30s in advance.
So the order of delegates is not known in advance for the next 100 blocks and instead only for the next block? After every block the next delegates is selected randomly?
Say there is only one delegate left from the 100 that has not yet been selected to find a block then he will be the one to find the next block?
-
Can the TPS of DPOS be as high as NXT Transparent Forging?
TPS = Transactions Per Second
Neither do I know whether NXT already implemented TF.
TF is not implemented completely yet. If DPOS gives a way to predict next miner with high probability then the answer is YES, I believe.
The exact order for delegates is known between every shuffle. So except for between shuffles you always know the next delegate with certainty at least 30s in advance.
So the order of delegates is not known in advance for the next 100 blocks and instead only for the next block? After every block the next delegates is selected randomly?
Say there is only one delegate left from the 100 that has not yet been selected to find a block then he will be the one to find the next block?
Every 100 blocks there is 1 shuffle that decides the next 100 blocks. So you always know on average 50 blocks in advance. Sometimes there is only 30 seconds and sometimes there is 3000 seconds depending upon where your delegate ended up in the shuffle.
-
This is an issue of trust once you get to 10K/sec because the bandwidth required is no longer able to be supported by a decentralized network. However, DPOS allows for centralization of delegates without losing the property of decentralized control.
+5%
This should be a major marking talking point.
-
Can the TPS of DPOS be as high as NXT Transparent Forging?
TPS = Transactions Per Second
Neither do I know whether NXT already implemented TF.
TF is not implemented completely yet. If DPOS gives a way to predict next miner with high probability then the answer is YES, I believe.
The exact order for delegates is known between every shuffle. So except for between shuffles you always know the next delegate with certainty at least 30s in advance.
So the order of delegates is not known in advance for the next 100 blocks and instead only for the next block? After every block the next delegates is selected randomly?
Say there is only one delegate left from the 100 that has not yet been selected to find a block then he will be the one to find the next block?
Every 100 blocks there is 1 shuffle that decides the next 100 blocks. So you always know on average 50 blocks in advance. Sometimes there is only 30 seconds and sometimes there is 3000 seconds depending upon where your delegate ended up in the shuffle.
Thanks.
So every delegate finds the same amount of blocks and gets the same reward as long as he is in the top 100, right?
Wouldn't shuffling after every block with the probability based on votes received make it harder to collude because a double spender would have to know the order of the next delegates?
-
So every delegate finds the same amount of blocks and gets the same reward as long as he is in the top 100, right?
No .. reward is fees-only .. so depends on the amount of transactions in the block .. but on 'average' all can get the same
-
Actually the network calculates the average fee per block and pays all delegates the moving average.
This is done to prevent perverse incentives where delegates attempt to steel trx from each other or ignore each others blocks.
Sent from my iPhone using Tapatalk (http://tapatalk.com/m?id=1)
-
What about this: Wouldn't shuffling after every block with the probability based on votes received make it harder to collude because a double spender would have to know the order of the next delegates?
-
What about this: Wouldn't shuffling after every block with the probability based on votes received make it harder to collude because a double spender would have to know the order of the next delegates?
That had a security issue where an attacker that could produce 2 blocks in a row would gain control of the random number sequence and thus produce all blocks forever.