Topic: [ANN][XMY] Myriad | Multi-Algo, Fair, Secure - page 68. (Read 850023 times)

Hey mister Myriadcoin, Look what they just released






 

Release Notes:
* Changes occur @ block 143,000
* 15 Second Blocks
* Scalability to match Visa TPS by 2021 & continued scaling through 2035 (See scalability projections here)
* Minimum TX & Relay Fee set to 1 DGB to prevent attacks
* MultiAlgo Difficulty adjustment code changed to make 51% attack much, much harder (See commit explanation)
* Transaction propagation changes based upon Microsoft research
* Reward adjusted to accommodate faster blocks, still stays the same (21 billion DGB in 21 years)
* OP_RETURN data increased to 80 bytes to allow embedding more data (like document hashes)
* Safe mode warning error fixed
* Check point added
* Private key compatibility issue fixed ( addresses from the DGB paper wallet address generator now work properly)

After much testing we are ready to get 4.0.0 into the wild! Please help spread the word and help get all mining pools & exchanges to update. As soon as an exchange or pool has updated please notify us and we will update our main list!


Big thanks to the Myriad devs!!

Seems that Qubit is just getting weaker and weaker. This is exactly the same thing that was happening to sha and scrypt before we switched to merge mining. The problem is each algorithm's hash rate is just too low which makes them susceptible to difficulty jumping attacks. This has also happened to groestl in the past but it seems groestl has recovered.

The question is, should we ignore it and wait for it to go away? We did this with groestl and it worked. It's the strength of Myriad... with 5 algos, if one has problems, the blockchain keeps moving.

Or should we switch to merge mining? We did this with sha and scrypt and it worked. Why not switch all algorithms to merge mining?  I guess the primary concern is that we would get dumped by DigiByte miners. Right now virtually all our miners are dumping. You can see there are days when Myriad gets about ~25USD of volume, spread across the exchanges, and it's all sells. That's about the USD value of all the MYR generated each day at market rate. If we switch to merge mining, I think it's safe to say the dumping isn't going to change.

Right now DigiByte could easily 51% attack us. They would not be able to get enough hash rate to control sha and scrypt, but they could attack us along groestl, skein, and qubit. One possible solution would be to add the yescrypt algo to Myriad and enable merge mining on all 6 algorithms. DigiByte miners do not mine yescrypt, and they probably do not even have the hardware for it (it's a CPU algo). With the addition of yescrypt, it would be extremely difficult for DigiByte to coordinate a 51% attack. On top of that, we would be able to switch to merge mining on all algorithms, effectively reducing the risk of difficulty jumping attacks to zero. And we would be able to include even more miners, upholding Myriad's reputation as "the coin for everyone."

While we're at it, why not add X11, cryptonite, blake, and anything else and merge mine all of them?

I don't see any problem with fragmented hash power? Myriad has always had a high hash rate relative to its market size and inflation rate. This is the predictable consequence of being a coin 'that anyone can mine'. More people in the world have hardware that can mine Myriad, than any other coin. 5 algorithms has always worked, and there's no evidence that 5 is too fragmented, or 6 either... 5 was basically a stab in the dark and it seems to work. For all we know 6 would work great, too.

As I see it, the multi-algoritm nature of MYR is a way of getting fastest feasible transaction times whilst mostly avoiding the difficulty-spiking and erratic transaction times of ZET, and the more independent-difficulty mining algorithm-groups there are then the better.  If someone is doing an algorithm changing hard-fork in future then please could people say what they think of the following :

Whilst it is unlikely^5 to get a difficulty spike in all five algorithms at the same time, that still could happen in the very rare worst case, so in selling our technology to telecomms engineers, who prefer to look at worst-case, performance could be improved somewhat if any more algorithms added have not been set up with difficulty independent of the other algorithms, but instead have been set up with difficulty anti-correlated to the instantaneous or most recent expectation of time-to-block-completion of the other algorithms.  That is, if all of the other algorithms are seeing falling hashrates since their difficulty was last calculated, then help things along by decrementing the difficulty of the next algorithm.  If there has been an increasing hashrate since D was last calculated then increment D of that next algorithm, and do so by formula acting on some pooled and public estimates of H so that not too many orphaned blocks result from changing D so often for this new algorithm.

i) rather than remove sha3 and swapping in an alternative algorithm, please could devs look at the feasibility of writing in a sixth algorithm?
That gains MYR by changing it to unlikely^6 to get a slow block.

ii) rather than replicating the independent D code for difficulty changing, please could devs look at feasibility of
 ii-a) anti-correlated D calculated per block, such that the 6th algorithm picks up blocks which nobody else will?
 ii-b) feasibility and networkwide requirements to change D of at least one algorithm more often than once per block.
         that needs a few safeguards, including that if somebody finds a block, there is an extra check that it satisfies the present D requirement as well as a previous D.

iii) my 32-bit PC's here could not install the -qt wallet.  Please could the readme clarify hardware requirements for installation of -qt wallets

iv) (a big ask) if one of the algorithms such as scrypt with sse2 support was blocking installation of -qt on some pc's then could the installer be modular so that it installs the wallet-qt with mining on several algorithms but not the sse2 scrypt ?  That is, less mining chances but at least it works.

What is the difference in this case? In both cases it seems to me grs-myr would be out and yescrypt in.
Because unless there's a very good evidence a feature is bad, pulling out stuff will cause the corresponding user base to be alienated and start a transition. In general features don't get removed 'just because'. If grs-myr is currently on FPGA then let the guys have their place; it is fine.

A good objection would be fragmenting the hash power too much. To that degree I would say we should unify everything that is not ASIC but that's an extreme consideration.

Related:
In the last few days I mined some grs-myr on P2Pool and enjoyed the massive payouts - over 161 MYR! Unfortunately, it seems I have pissed off the constant ~1MHs user who typically got 490 MYRs out of each block and now it's gone. It is unlikely I will mine in the next few days and so far P2Pool has not found a block in days. Shall p2pool not find a block soon, I encourage someone to point some mining there.

Why?

Not dropping, just swapping. I thought swapping myr-groestl was a bit more appealing because it would bring myriad into a bit more alignment with unitus.

grs-myr is chained too.
Considering the original idea of qubit was to be the CPU algo, swapping it for yescrypt might sound like a big deal... except qubit is currently the go-to algo for many GPU users.

What is the situation with skein BTW? A few months ago it still needed its own miner and wasn't integrated in sgminer yet. This makes the transition a bit problematic.

I don't think dropping algorithms is good btw.

FPGA implementations might not be good for Myriad distribution, but as you said yourself, it's not just distribution that matters, it's *fair* distribution. FPGAs are fair play and any algo is subject to them if it becomes sufficiently popular. I don't think it's fair to switch algos just because of FPGAs.

Qubit is the weakest of the algos because it is a chained algo. Even 8bitcoder said himself long ago when Myriad was released that Qubit could be swapped for something else.

After discussing it on IRC and thinking about it a bit, maybe it would be better to swap out myr-groestl for yescrypt. Then we'd be more closely aligned with Unitus, which uses qubit, yescrypt, and skein (but not myr-groestl), and I've also heard there is an fpga implementation of myr-groeslt.

What about swapping out qubit for yescrypt? Saw some discussion on IRC about it.

Looks like someone is jumping on with a bunch of hash when diff gets low, see below:

Then jumps off when diff goes up, and then qubit only finds a couple blocks in next few hours. A lot of those blocks above where mined by an address starting with MJJx...

Thanks HashEngineering 

Why isn't qubit finding its fair share of blocks?

I am working on the code for the Android App.  A trusted peer won't help until the app is updated for the recent forks.

EDIT:  The app has been updated on Google Play.  The source code and downloads have been updated on GitHub (https://github.com/HashEngineering/myriadcoin-wallet/releases/tag/v0.9.2.16).

Feel free to send some coins:  
BTC:  12vd4aST1F61Czh9EvcUcQxJggdJyKBbNx
MYR:  MSPBwfcruZNxBXfDEvFv2vu4MXX3ccXaNe

Who was update on 0.9.2.16?

Myr-groestl Hawaii binary (290, 290x, ...), 63 Mh/s on r9 290x @1100/150, compatible with stock miner.
PM for info.

Help to win MYR voting on Cryptopia -> https://www.cryptopia.co.nz/Home/Voting
 

Can anyone weigh in on this comment? https://github.com/myriadcoin/myriadcoin/commit/15a341c28dfdcfab90546dd3a2962439b8ee907d#commitcomment-12124053

I'm pretty sure that when you upgrade to v0.9.2.16 you need to launch the daemon or qt with the -reindex flag, otherwise it will crash (although deleting the datadir also works, as you discovered).

What's altcoin ping-ping?

PM your MYR address if you want to test your wallet.

more than 6GB free [its an i5 running 64-bit]

synch completed from 1yr 36wk behind up to present 07-nov-2015 in about 2h, NOTE but not on the first attempt.

A well-known altcoin exchange has still not sent my test transaction to it a couple of hours after I'd ok'd their email confirmation and it really is not yet outbound from the exchange.  Perhaps I'll see the daily merge-mine byproduct from multipool.us overnight, but until then my myriadcoin-qt and wallet is at "it compiles, it synchs, it is not yet fully tested".

Anyone up for a game of altcoin ping-ping tomorrow to test basic function of the MYR currency ?