And the number of iterations must be low enough so that validation of blocks becomes fast enough.
Seems you never heard of asymmetric PoWs, where only proof attempts require a lot of memory, but verification requires negligible... But is it really a problem if validation requires an equal amount of memory? It's a big problem in SPV clients, e.g. on mobile devices, and also makes the network more susceptible to ddos-attacks with false proofs. Hmm... Ok. Darn. I think my random index jumping would work, but it does require that the verification algorithm has the same amount of memory. I need to go back to the drawing board I guess. There are already PoWs that combine a large memory requirement with instant verifiability (and where memory access dominates computation). |
|
|
|
And the number of iterations must be low enough so that validation of blocks becomes fast enough.
Seems you never heard of asymmetric PoWs, where only proof attempts require a lot of memory, but verification requires negligible... But is it really a problem if validation requires an equal amount of memory? It's a big problem in SPV clients, e.g. on mobile devices, and also makes the network more susceptible to ddos-attacks with false proofs. |
|
|
|
And the number of iterations must be low enough so that validation of blocks becomes fast enough.
Seems you never heard of asymmetric PoWs, where only proof attempts require a lot of memory, but verification requires negligible... |
|
|
|
Use the search button.. a lot of people have tried to do this in various ways.. All of them have failed.
They failed because they were computation dominated. For a PoW that spends a minority of runtime on computation, and a majority on random memory access latency, commodity DRAM could be the ASIC. |
|
|
|
|
Cuckoo Cycle is the only algorithm that's more memory-intensive than compute-intensive
(computation taking only 33% of its runtime).
It can furthermore require 100s of MBs, while remaining instantly verifiable.
There is an unclaimed bounty of $1000 for a GPU implementation that can match a server CPU
in performance.
|
|
|
|
Just curious if anyone has any insights into what is the best coin and or algo for CPU mining.
Cuckoo Cycle is designed to be ideal for CPU mining by minimizing computation and maximizing random memory access latency. It uses 3 orders of magnitude more memory than scrypt while being instantly verifiable. See https://github.com/tromp/cuckoo for details... |
|
|
|
(anybody remember Aurocoin?)
Even its name is being forgotten:-) |
|
|
|
Monero is currently under attack with a spam attack, while no other CN coin seems to be. The attacker is spamming tx with random, large mixins (initially only mixin 101 was used, until I called them out on it in the official thread).
And what is the problem with these txs? Are they invalid? Does it take too much effort to decide their validity? If so, couldn't you require that all tx come with a small and instantly verifiable proof-of-work, whose difficulty scales with tx validity checking effort? |
|
|
|
So one attack vector is forging with timestamps artifiically advanced by almost 15 secs,
and having many nodes disagree on whether the new block is valid or not...
You need coins to forge, a lot of them, that's why it's called Proof-of-Stake  Why would you destroy your own investment if you spent billions of USD to buy up 51% of coins (or 90% of coins when TF is implemented)? I only need a small fraction of the active stake to forge the occasional block. My question remains, what happens when many nodes disagree on whether some block is valid? |
|
|
|
My understanding is that PoS has attack vectors that are not present in PoW, giving the attacker a cheaper way to totally control/destroy the network.
Can you elaborate? I have observed exchange price action and noticed that when 1% of NXT is purchased at market price, this drives the price up 25%. As I pointed out in the other bitcointalk thread, my basic NXT PoS soundness questions raised in https://nxtforum.org/general/forging-questions/remain unanswered. So one attack vector is forging with timestamps artifiically advanced by almost 15 secs, and having many nodes disagree on whether the new block is valid or not... |
|
|
|
I recently tried to lay to rest some concerns I have about the technical underpinnings of the PoS system as used in NXT. See this thread on the NXT forum: https://nxtforum.org/general/forging-questions/Needless to say, my concerns have not been laid to rest. Furthermore, I have to say that the lack of technical expertise in the NXT community is somewhat worrying. You'll just have to search Bitcointalk original NXT thread (the one that's over 1000 pages long) for answers, developers are busy coding new features like crazy, no time to answer everyone's questions I think NXT is in a deplorable state if basic questions about achieving consensus (the single goal of PoS) cannot be answered in a straightforward manner. Such knowledge should not be limited to core developers or buried deep in long discussion threads. It should be very prominent in FAQs and wikis and whitepapers. It is not. That is cause for worry... |
|
|
|
I recently tried to lay to rest some concerns I have about the technical underpinnings of the PoS system as used in NXT. See this thread on the NXT forum: https://nxtforum.org/general/forging-questions/Needless to say, my concerns have not been laid to rest. Furthermore, I have to say that the lack of technical expertise in the NXT community is somewhat worrying. |
|
|
|
only thing different about our project is that it is cheaper to invest.
Unlike Ethereum sales price of Ethe will be cheaper. After long discussion among our team, we have decided to sell one Ethe for 0.00001 BTC( 1000 sat).
Brilliant! You inspired me to launch Ethereos, which will be a cheaper version of Ethereal. Ethereos uses Eth rather than Ethe. That way you save even more characters talking about it. After long discussion with myself, I have decided to sell one Eth for 0.00000001 BTC (1sat). Beat that! lol. I wonder what the name of the lead dev is: Vitalicorice? The lead dev is Zuckers Galore |
|
|
|
only thing different about our project is that it is cheaper to invest.
Unlike Ethereum sales price of Ethe will be cheaper. After long discussion among our team, we have decided to sell one Ethe for 0.00001 BTC( 1000 sat).
Brilliant! You inspired me to launch Ethereos, which will be a cheaper version of Ethereal. Ethereos uses Eth rather than Ethe. That way you save even more characters talking about it. After long discussion with myself, I have decided to sell one Eth for 0.00000001 BTC (1sat). Beat that! To secure your share of Eth you need to send BTC to this address: 1down9the7toilet3Rw2vuPwbnStIwpAUX5a |
|
|
|
Why don't you look at the documentation on GPU programming?
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
My Cuckoo Cycle PoW spends 67% of its time on memory access. In other words, if you leave out all memory accesses, and just do the remaining computations, then runtime reduces to 33% of original. That's what I would call "memory intensive". You can do the same test for NeoScypt. Run it with all memory accesses removed, and see how much runtime is reduced. |
|
|
|
The full sentence in the whitepaper is: "These constraints were supposed to protect hash from GPU and ASIC implementation, but a GPU miner appeared on the scene in 2 weeks after this technology got public attention." Thus, contextually we know that his meaning in the word "protect" is "ensure they do not exist". He considers the very existence of a GPU miner a failure of the algorithm, when, in fact, a GPU miner can and should exist as long as it the performance gap is closed. Currently GPU miners are 2-3x as performant / efficient as CPU miners, and by dga's calculations they shan't exceed ~5x the performance / efficiency. Thus the algorithm has completely succeeded at what it purports to do, and has met its primary goal.
If we're nitpicking, then closing the gap would mean that GPUs are no faster than (the fastest) CPU. A 2x-3x gap is a narrowed, not a closed gap. |
|
|
|
Additionally while we do yet have an idea of the cap, we do know that the total number will always go up; however, as designed, it will go down as a percentage of issued Ether. This is expected to eventually reach the same amount of Ether lost on an annual basis resulting from loss, theft, or fraud, which in time, will create a natural equilibrium.
Ether only get lost from loss (e.g. forgotten password, sent to bad address), not from theft or fraud, which is just irregular change of ownership... |
|
|
|
Man...i have just opened the Terms and Conditions page :
"....for technical reasons EthSuisse may need to change the annual rate of new creation to a value lower than 26% to account for changes to the Ethereum mining algorithm or to address other issues that may arise. There is no guarantee that this percentage will be accurate or that this rate will continue at the same level....."
For FIAT, a 25% inflation rate would spell disaster. But not even 26% is guaranteed. Unbelievable !!!!
26% of ETH sold in the IPO is the guaranteed maximum. Since the yearly inflation is a fixed amount of ETH, it is a decreasing percentage as the years go by, and will at some point become less than even the few % of ETH that inevitably gets lost every year. |
|
|
|
What I find interesting:
With 1 BTC you could buy about 0,1-1% of coins like dogecoin,darkcoin in their beginings.
With 1 Btc here you ar buying about 0,0000385%
Check your math: It's a 0.0000385 fraction which is 0.00385 % |
|
|
|
Well peoples are too lazy to make a new hash for cryptocoins.
Speak for yourself; I'm busy working on Cuckoo Cycle, a new PoW (not a hash). |
|
|
|
|
Show Posts
