Altcoins and Scamcoins - Know the difference.

[snailbrain]

Distributed p2p domain namespace(?) system:
Namecoin -- 2012(?) 2011 - Oldest Altcoin?

[Invest0r]

BitBar - Scrypt - Hybrid proof-of-work/proof-of-stake and a fast, continuous difficulty and reward calculation.

[ecliptic]

Distributed p2p domain system:
Namecoin -- 2012(?) 2011 - Oldest Altcoin?

indeed, i found some more reading on github, updated the dates and order and added the link.

i also dropped the premined coin sections - frankly there's too many damn coins to list there.  the very few ones that were novel AND premined are listed in their novel section with a disclaimer they were premined and thus dead.

BitBar - Scrypt - Hybrid proof-of-work/proof-of-stake and a fast, continuous difficulty and reward calculation.

Scrypt and hybrid PoW/PoS were already implemented prior to Bitbar, i will look if there is a novel method in their diff/reward calc

[WindMaster]

However I think that scrypt() idea is fundamentally bad.  It sounds like it will be another litecoin scrypt() cgminer gpu catastrophe.

I would not be surprised if people, maybe even many people independently, have forked and created their own cgminer in secret which is capable of utilizing GPUs, giving them several order of magnitude advantage over everyone else.

or am i wrong?  even the devs seem to say it is possible.  simply saying "well the code doesn't work right now" is not going to prevent someone from doing it.  and it's much, much faster.  if the coin has/had any worth it will be done

Given that the exact hashing algorithm (substitution of chacha20/8 in place of scrypt20/8, and Kekkac512 in place of SHA256) was a surprise to everyone at the coin's launch, there was a period of time where CPU mining would have had to be dominant.  The lion's share of YAC was mined during the first day by CPU's.  The author of Reaper (whose OpenCL kernel for mining Litecoin was later lifted for cgminer, BTW, so cgminer wasn't the original), mtrlt, reports needing about 13.5 hours of work to get Reaper going with any sort of hash rate that had a noteworthy advantage, and he also reports having had a late start.  Probably few people would've been able to implement it faster than the person who wrote the OpenCL kernel just about everyone mining scrypt coins with GPU's is using.  So, it's probably reasonable to assume that GPU mining of YAC has a high probability of having been entirely CPU for the first 13.5 hours, with a fair probability it was around day 2 before anyone had a GPU implementation up and running (given that most other people attempting it likely would've taken longer than mtrlt).  I did it in 8 hours but ended up with hash rates on a 6950 not much above my dual Xeon E5450 servers, so I didn't bother pursuing GPU mining YAC further.  It appears low hash rate results were common among the people that did attempt OpenCL implementations, as most of us aren't as good at optimizing OpenCL in "thinking outside the box" ways like mtrlt.  So, just my humble guess is that GPU implementations probably started occurring around day 2, with many poorly optimized versions not giving much advantage over CPU's, so I'd consider those a wash.  Until people speak up with more data, we don't know for sure that anyone other than mtrlt managed to achieve significant hash rates for GPU mining of YAC.  There could be, and it's also possible there weren't.  Time will tell, probably.

I've heard it's possible and that at least one forum user had a working implementation after a couple weeks, but AFAIK all of the 23 days since the coin's creation, save for the first hours (and by several people's experiences CPU mining the first 8 hours, it didn't seem like there was anyone GPU mining at that point), it's been more profitable to GPU mine about any other GPU coin rather than YAC.

My GPU's are happily mining LTC, if anyone is curious.  And I'm one of the few that have admitted to having a GPU implementation of scrypt+chacha20/8+Keccak512(N,1,1), albeit producing nowhere near the hash rates that mtrlt, the author of Reaper, reported.

As I understand it, if YAC's price rises there may a window of GPU mining being profitable, till N increases so much it becomes too memory heavy for GPUs again. I don't have the theoretical knowledge to assess this, though.

Incidentally, exchange price has remained so low for the whole first month that anyone could buy easily as many as the top hasher can mine

+1, I've bought far more YAC than I mined.  And I seem to be someone people point at as having mined a lot of YAC (I'm not so sure, I've seen in the YAC block explorer what the people who were CPU mining right from the coin's launch raked in).

[ecliptic]

However I think that scrypt() idea is fundamentally bad.  It sounds like it will be another litecoin scrypt() cgminer gpu catastrophe.

I would not be surprised if people, maybe even many people independently, have forked and created their own cgminer in secret which is capable of utilizing GPUs, giving them several order of magnitude advantage over everyone else.

or am i wrong?  even the devs seem to say it is possible.  simply saying "well the code doesn't work right now" is not going to prevent someone from doing it.  and it's much, much faster.  if the coin has/had any worth it will be done

Given that the exact hashing algorithm (substitution of chacha20/8 in place of scrypt20/8, and Kekkac512 in place of SHA256) was a surprise to everyone at the coin's launch, there was a period of time where CPU mining would have had to be dominant.  The lion's share of YAC was mined during the first day by CPU's.  The author of Reaper (whose OpenCL kernel for mining Litecoin was later lifted for cgminer, BTW, so cgminer wasn't the original), mtrlt, reports needing about 13.5 hours of work to get Reaper going with any sort of hash rate that had a noteworthy advantage, and he also reports having had a late start.  Probably few people would've been able to implement it faster than the person who wrote the OpenCL kernel just about everyone mining scrypt coins with GPU's is using.  So, it's probably reasonable to assume that GPU mining of YAC has a high probability of having been entirely CPU for the first 13.5 hours, with a fair probability it was around day 2 before anyone had a GPU implementation up and running (given that most other people attempting it likely would've taken longer than mtrlt).  I did it in 8 hours but ended up with hash rates on a 6950 not much above my dual Xeon E5450 servers, so I didn't bother pursuing GPU mining YAC further.  It appears low hash rate results were common among the people that did attempt OpenCL implementations, as most of us aren't as good at optimizing OpenCL in "thinking outside the box" ways like mtrlt.  So, just my humble guess is that GPU implementations probably started occurring around day 2, with many poorly optimized versions not giving much advantage over CPU's, so I'd consider those a wash.  Until people speak up with more data, we don't know for sure that anyone other than mtrlt managed to achieve significant hash rates for GPU mining of YAC.  There could be, and it's also possible there weren't.  Time will tell, probably.

I've heard it's possible and that at least one forum user had a working implementation after a couple weeks, but AFAIK all of the 23 days since the coin's creation, save for the first hours (and by several people's experiences CPU mining the first 8 hours, it didn't seem like there was anyone GPU mining at that point), it's been more profitable to GPU mine about any other GPU coin rather than YAC.

My GPU's are happily mining LTC, if anyone is curious.  And I'm one of the few that have admitted to having a GPU implementation of scrypt+chacha20/8+Keccak512(N,1,1), albeit producing nowhere near the hash rates that mtrlt, the author of Reaper, reported.

As I understand it, if YAC's price rises there may a window of GPU mining being profitable, till N increases so much it becomes too memory heavy for GPUs again. I don't have the theoretical knowledge to assess this, though.

Incidentally, exchange price has remained so low for the whole first month that anyone could buy easily as many as the top hasher can mine

+1, I've bought far more YAC than I mined.  And I seem to be someone people point at as having mined a lot of YAC (I'm not so sure, I've seen in the YAC block explorer what the people who were CPU mining right from the coin's launch raked in).

I see

it's not the at launch/early hashrate i was concerned about, i am more thinking on the long term, gpu mining for many days, weeks, months while everyone else is on cpus.

but maybe it is still quite resistant to gpus being that great even if you can get them to work

[sor.rge]

I'm not so sure, I've seen in the YAC block explorer what the people who were CPU mining right from the coin's launch raked in

You seem to completely disregard the likely possibility that the author had an optimized GPU miner before the launch.

[WindMaster]

I'm not so sure, I've seen in the YAC block explorer what the people who were CPU mining right from the coin's launch raked in

You seem to completely disregard the likely possibility that the author had an optimized GPU miner before the launch.

Yes, that is possible.  And I was one of the first people weighing in on the likelihood that the original developer's claims that YAC was GPU-resistant were likely not correct, back on the first day of the coin's launch (along with TacoTime, who was saying the same thing).  Unfortunately, we're not going to know for sure, especially since the original developer pulled the usual Houdini disappearing act after the coin launched.

I've examined the first ~100 blocks in the YAC block explorer trying to determine any of the original developer's coinbase addresses to try to figure out a lower bound for how much he made.  I didn't find anything exciting in the first 100 blocks that indicated much more than just CPU mining (so I'll certainly say CPU instamining by the original developer was a possibility).  It's entirely possible the original developer did have a GPU implementation but waited before turning it up.  There's no way to tell at this point.

A lot of people were real quick to turn up massive numbers of Amazon AWS instances and mine with them.  It's difficult to distinguish between those people and any hypothetical GPU's or GPU farms while examining blocks with the YAC block exporer.  ~9 to 9.5 hours after the coin launched, I had tons of Amazon instances mining too.

[WindMaster]

I see

it's not the at launch/early hashrate i was concerned about, i am more thinking on the long term, gpu mining for many days, weeks, months while everyone else is on cpus.

but maybe it is still quite resistant to gpus being that great even if you can get them to work

This part we don't know yet.  At least the developer of Reaper (mtrlt) reported decreases in GPU hash rate roughly proportional to decreases in everyone else's CPU hash rates as N started to increase.  It's a little too early to tell what the situation will be when N starts getting high and both GPU's and CPU's are increasingly falling back to external memory (since the TMTO related to lookup gap only improves things to a point, which is why everyone gets faster hash rates on Litecoin with a lookup gap of 2 rather than 4, for example).

Future prevalence of GPU mining of YAC is probably going to depend on what exchange rates do, and whether N rises enough to whack GPU hash rates (for anyone that actually did succeed in making a good implementation with decent hash rates) while it is still more profitable to mine other coins with GPU's.  Small arms race, perhaps.  Right now, I'd bet on buying YAC cheap on Bter if someone is betting YAC will rise significantly in price later on, rather than tying up GPU's that could be mining Litecoin instead.

I only implemented OpenCL for YAC at N=32, and haven't bothered trying for other N values.  So, unfortunately I don't have good benchmark data for falloff of GPU hash rates as N increases.

I've gone on record several times in saying that YAC's N started too low at launch though.  I would've started N significantly higher than 32.  As a result, I pulled off an FPGA implementation that performed rather well at N=32, but became iffy at N=64 and couldn't be placed'n'routed by the Xilinx tools at all for N=128.

[ecliptic]

I see

it's not the at launch/early hashrate i was concerned about, i am more thinking on the long term, gpu mining for many days, weeks, months while everyone else is on cpus.

but maybe it is still quite resistant to gpus being that great even if you can get them to work

This part we don't know yet.  At least the developer of Reaper (mtrlt) reported decreases in GPU hash rate roughly proportional to decreases in everyone else's CPU hash rates as N started to increase.  It's a little too early to tell what the situation will be when N starts getting high and both GPU's and CPU's are increasingly falling back to external memory (since the TMTO related to lookup gap only improves things to a point, which is why everyone gets faster hash rates on Litecoin with a lookup gap of 2 rather than 4, for example).

Future prevalence of GPU mining of YAC is probably going to depend on what exchange rates do, and whether N rises enough to whack GPU hash rates (for anyone that actually did succeed in making a good implementation with decent hash rates) while it is still more profitable to mine other coins with GPU's.  Small arms race, perhaps.  Right now, I'd bet on buying YAC cheap on Bter if someone is betting YAC will rise significantly in price later on, rather than tying up GPU's that could be mining Litecoin instead.

I only implemented OpenCL for YAC at N=32, and haven't bothered trying for other N values.  So, unfortunately I don't have good benchmark data for falloff of GPU hash rates as N increases.

I've gone on record several times in saying that YAC's N started too low at launch though.  I would've started N significantly higher than 32.  As a result, I pulled off an FPGA implementation that performed rather well at N=32, but became iffy at N=64 and couldn't be placed'n'routed by the Xilinx tools at all for N=128.

You were able to make an FPGA design for Scrypt(N=32,64)?  How did it work?

[ecliptic]

Generally Bad Items(tm) Which You Shouldn't Do

* Centralize and/or close source anything. Cyrptocurrency means P2P and open source

I agree with you on everything except sentences above. All important and mission-critical software and hardware on this planet is centralized and closed source.

The majority of servers and supercomputers in existence run Free Open Source Software.

Open source software is more secure than closed source software

[digitalindustry]

Coins which have actually tried something fundamentally new, for better or worse.  In order of their introduction.


Coins which use a novel POW:

Bitcoin - SHA256 originator
Litecoin - Scrypt originator
PPCoin - Proof of Stake

Distributed p2p domain system:

Namecoin

Coins which have used ascending rewards to mitigate premining/early mining of the network:

Nibble
Digicoin

+1

Great thread 

+ 1  i will say i did look into NVC it has been quite innovative , but terribly marketed. but its also on BTC-e I think its a stay.

[casascius]

True and false religion: how to tell the difference

True religion: the one I believe in
False religion: all others
True prophet: The guy I listen to
False prophet: all the guys who say they are true prophets but are lying, except for the one I believe in
God: the guy who started my religion and talks to my prophet
Satan: the guy who tries to get people to leave my religion and join others
Good: doing whatever my religion says to do
Evil: doing whatever my religion says not to do, including believing in other religions


Etc.

And so it follows:

Altcoin: the one I hold
Scamcoin: the ones I don't

[Impaler]

Some additional innovations that FRC has

* Smoothly declining block rewards that eliminate the disruptive volatility inducing reward-halving events of all other coins
* Optimized difficulty adjustment parameters, adjusts every 9 blocks up or down 5% based on 144 block rolling window (will go live in a day or two through hard fork).  These parameters were identified through an exhaustive simulation process to be ideal for FRC and similar coins with 10 minute block intervals like BTC and TRC.

Also I don't think your characterization of '80% for developers' is correctly describing FRC, developers have pledged to give away all these coins through a non-profit foundation not keep them for themselves.

[ecliptic]

Some additional innovations that FRC has

* Smoothly declining block rewards that eliminate the disruptive volatility inducing reward-halving events of all other coins
* Optimized difficulty adjustment parameters, adjusts every 9 blocks up or down 5% based on 144 block rolling window (will go live in a day or two through hard fork).  These parameters were identified through an exhaustive simulation process to be ideal for FRC and similar coins with 10 minute block intervals like BTC and TRC.

Also I don't think your characterization of '80% for developers' is correctly describing FRC, developers have pledged to give away all these coins through a non-profit foundation not keep them for themselves.

True enough, i didn't know the specifics of FRC.

[shakezula]

I don't think your characterization of '80% for developers' is correctly describing FRC, developers have pledged to give away all these coins through a non-profit foundation not keep them for themselves.

  You can give away sand in the desert too...

[MashRinx]

Which bounties though?  I am not as familar with them

Any premine is BS, regardless of the stated reason.  I have seen more than one coin released recently with at least 1,000,000 coins listed as a premine for "bounties", to pay the "foundation" (really?) etc.  Those who would defend this will do so for multiple reasons, but it all boils down to defending their holdings in the given coin.

There have been at least a couple coins recently that have tried to reduce the amount of early/surprise mining that I appreciate.  Others may disagree...

[xan_The_Dragon]

i agre with all of this

[GSnak]

A lot of this stuff sounds like actual coding and not just "Find and Replace". I don't think much of it will get done as it sounds like work.

[GSnak]

Supercomputers just like battleships, nuclear reactors and so on run customized software that maybe originated from Open Source software. Anyone running stock Apache, PHP,
MySQL, GNU/LINUX or whatever else is an idiot so I don't count those cases as good example of Open Source software.

Hmm. 61.3% of server admins are idiots. Sad, sad world we live in.

[nprussell]

Excellent posts OP. This thread needs stickying.