[LTC] Changing the litecoin Proof of Work function to avoid ASIC mining?

[yochdog]

I am concerned by these developments, in light of LTC's initial purpose. 

Is anyone in touch with the devs?  Is there a way to get ahold of them for discussion?

Perhaps they have been bought off with some early pre-orders? 

[hrstuffnstuff]

I think people really miss the big picture here. If you are worried about centralization then trying to push ASICs out with memory changes will not help. This is why: the first time you change the code, ASICs will not work. People who develop ASICs will immediately get wise and as soon as the technology catches up(and it will) they will make ASICs and just mine with them. If they are smart they will mine to different pools under different names, etc. The only reason you know there are ASICs on the market is because people tell you. If no one told you, all you would see is a spike in difficulty over time. Even that can be spread out to not look conspicuous. Now you have one company doing untold percentage of the network hashrate and you have centralization all over again. This time it's without public knowledge of it which is even more dangerous.

Is centralization bad? yes. However, ASICs don't cause centralization by themselves. If people in the LTC community got more group buys, and got more ASICs into people's hands then there wouldn't be a centralization issue at all. It is only because all the GPU miners have no desire to get ASICs or be a part of the arms race (and who can blame them) that centralization happens.

[BrewCrewFan]

I think people really miss the big picture here. If you are worried about centralization then trying to push ASICs out with memory changes will not help. This is why: the first time you change the code, ASICs will not work. People who develop ASICs will immediately get wise and as soon as the technology catches up(and it will) they will make ASICs and just mine with them. If they are smart they will mine to different pools under different names, etc. The only reason you know there are ASICs on the market is because people tell you. If no one told you, all you would see is a spike in difficulty over time. Even that can be spread out to not look conspicuous. Now you have one company doing untold percentage of the network hashrate and you have centralization all over again. This time it's without public knowledge of it which is even more dangerous.

Is centralization bad? yes. However, ASICs don't cause centralization by themselves. If people in the LTC community got more group buys, and got more ASICs into people's hands then there wouldn't be a centralization issue at all. It is only because all the GPU miners have no desire to get ASICs or be a part of the arms race (and who can blame them) that centralization happens.

Well its the price point. That is what the issue is. Most people have a hard time getting 500 bucks together for another video card.... now with these asics your gonna ask to get 2K+? The only ones that will have this are those with money, and then what happens is these people with these "asics" will then dump coins on the market even faster and lower the prices even more, which in turn kills off even more normal users due to less profits. By the time something is "more affordable" it does not make sense to spend the money because the ROI is so damn low.

[mr.n00blar]

Save the clocktower, vote yes to resisting ASICs.

[hrstuffnstuff]

Well its the price point. That is what the issue is. Most people have a hard time getting 500 bucks together for another video card.... now with these asics your gonna ask to get 2K+? The only ones that will have this are those with money, and then what happens is these people with these "asics" will then dump coins on the market even faster and lower the prices even more, which in turn kills off even more normal users due to less profits. By the time something is "more affordable" it does not make sense to spend the money because the ROI is so damn low.

I agree on all your points, unfortunately changing the POW doesn't fix this, it only makes it worse.

[hrstuffnstuff]

I've seen them on IRC. In particular, ceblee was saying not too long ago that LTC is a coin for the people because it can be mined on consumer hardware. Now he says, and I quote directly, "This isn't a democracy."

How do you go from, "coin for the people" to "I RUN THIS SHIT"?

That is little out of context, he went on to say that a democracy of uninformed people does not work. What he was referring to was the fact that people don't understand the dangers of changing the POW and he does. He also said in that same chat that if you can convince people at the litecoin foundation with arguments that the risk is worth it, he is happy to listen and possibly take action if the dangers are understood. At least that is how it read to me.

[User705]

I am concerned by these developments, in light of LTC's initial purpose. 

Is anyone in touch with the devs?  Is there a way to get ahold of them for discussion?

Perhaps they have been bought off with some early pre-orders? 

I've seen them on IRC. In particular, ceblee was saying not too long ago that LTC is a coin for the people because it can be mined on consumer hardware. Now he says, and I quote directly, "This isn't a democracy."

How do you go from, "coin for the people" to "I RUN THIS SHIT"?

You can always vote with your feet and just sell your LTC.

[tromp]

I think people really miss the big picture here. If you are worried about centralization then trying to push ASICs out with memory changes will not help. This is why: the first time you change the code, ASICs will not work. People who develop ASICs will immediately get wise and as soon as the technology catches up(and it will)

No; ASIC technology will never catch up to requiring many GBs of memory.

Just adopt a PoW that takes at least a GB of memory to solve
(with no time-memory tradeoff) and that depends on latency rather than bandwidth.

That will put an end to specialized mining hardware.

[hrstuffnstuff]

No; ASIC technology will never catch up to requiring many GBs of memory.

Just adopt a PoW that takes at least a GB of memory to solve
(with no time-memory tradeoff) and that depends on latency rather than bandwidth.

That will put an end to specialized mining hardware.

Yes, ASIC technology absolutely will catch up. It may take time but considering how fast larger memory sizes come out and then cheapen it should be no surprise that it will. I suggest you read up on Moore's law and the new memory technologies coming out. This is and always will be a function of price of a coin and price of memory (and of course difficulty).

[tromp]

No; ASIC technology will never catch up to requiring many GBs of memory.

Just adopt a PoW that takes at least a GB of memory to solve
(with no time-memory tradeoff) and that depends on latency rather than bandwidth.

That will put an end to specialized mining hardware.

Yes, ASIC technology absolutely will catch up. It may take time but considering how fast larger memory sizes come out and then cheapen it should be no surprise that it will. I suggest you read up on Moore's law and the new memory technologies coming out. This is and always will be a function of price of a coin and price of memory (and of course difficulty).

I suggest you read up on http://en.wikipedia.org/wiki/Random-access_memory#Memory_wall

Main memory latencies have improved very slowly over time.
Highly optimized ASICs for latency-hard algorithms already exist.

They're called memory chips.

[peterlustig]

WTF 2/3 of people support a PoW change. Maybe the devs of LTC should have thought of ASIC resistance before setting the n-Factor and memory requirements absurdly low, but lying to everyone how ASICS and GPUs are an impossibility.

[deamon]

Yes. And time is running out. One guy on the middlecoin pool already has 800MHash/sec ASICs online, and stated that he will up it to 4000MHash soon. That is several percent of the whole LTC network. Action is required now, before the voting power of the GPU miners is lost!

++1

Are there serious plans out to restrict ASICS?
Or are this only all speculative, so no one will really ban the ASICS?
Eventually are the ASICS the "backbone" of the coins, but its a game with the fire?

[tromp]

No; ASIC technology will never catch up to requiring many GBs of memory.

Just adopt a PoW that takes at least a GB of memory to solve
(with no time-memory tradeoff) and that depends on latency rather than bandwidth.

That will put an end to specialized mining hardware.

Yes, ASIC technology absolutely will catch up. It may take time but considering how fast larger memory sizes come out and then cheapen it should be no surprise that it will. I suggest you read up on Moore's law and the new memory technologies coming out. This is and always will be a function of price of a coin and price of memory (and of course difficulty).

I suggest you read up on http://en.wikipedia.org/wiki/Random-access_memory#Memory_wall

Main memory latencies have improved very slowly over time.
Highly optimized ASICs for latency-hard algorithms already exist.

They're called memory chips.

There will pretty much always be a time-memory tradeoff, because you can always avoid using memory by recomputing the data when you need it...

There is no TMTO if recomputing the data to save half the memory is millions of times slower, as it is with the right memory-hard PoW.

[j23a]

Yes, 100%

It's such a scam, the only people who win are the manufactures, who take pre orders months in advance, and sell it to the people who pre ordered them just when it's more profitable for the ASIC companies to sell the ASICs rather than to mine with them. The difficulty goes way up, ruining it for the people who bought the ASICs and us regular folks. And on top of that, the ASICs have no resale value, unlike the GPUs.

I rather just spend the 8 thousand dollars on Litecoins and see if it goes up in the months before the ASIC is released, and the many months, maybe over year to get a ROI.

In fact, I rather buy 25Mh of GPUs than buy a 25Mh ASIC, because at least I know I can resell the GPUs in a couple of months and break even, rather than wait a couple of months to get the ASICs and then wait who knows how many months to break even, if ever. And then have a dead weight sitting there.

[Sophokles]

Yes. And time is running out. One guy on the middlecoin pool already has 800MHash/sec ASICs online, and stated that he will up it to 4000MHash soon. That is several percent of the whole LTC network. Action is required now, before the voting power of the GPU miners is lost!

++1

Are there serious plans out to restrict ASICS?
Or are this only all speculative, so no one will really ban the ASICS?
Eventually are the ASICS the "backbone" of the coins, but its a game with the fire?

Litecoin was designed to be ASIC proof. Even GPU proof, but that did not work out either. ASIC is definitely not the back bone of Litecoin yet. But changing the PoW function is hazardous, certainly. Still, I vote for it. Because Litecoin will lose one of its distinguishing features if it becomes dominated by the ASIC miners.

[hellscabane]

Although the votes are skewed towards yes, I'm glad a significant amount of talk is for not changing it.

As mentioned earlier, doing so would lead to a slippery slope; and is just a means of delaying an "inevitability." The issue is that the premise was already shaky from the start; in such a burgeoning and young area (in technology, no less) a sweeping statement saying that a protocol would be ASIC-Proof was a stretch in itself. And the trickiest thing is how to ensure that the protocol remains ASIC-Proof. If we change it once, and then realize that we were wrong, we'll have to do it again. And again. And again...

I think that the implications of changing are far greater, and more far reaching than leaving it as is. How people react to it is another story (for all we know it could bolster the value of the network), but is it worth the "risk?" It would be much simpler changing the mission statement of LTC in my opinion. Once again though, the reaction to this type of change is unknown. Plus you lose the main "differentiating" feature of LTC. But for me, that is much more worth the "risk" than changing the protocol (and possibly changing it again and again).

I think the loss in credibility of possibly having to change a protocol multiple times far outweighs the credibility loss in changing a relatively young mission statement.

[greentea]

once these ASICs come out, they're going directly to the profit-switching pools like multipool/middlecoin.

so glad Quark doesnt have this problem ... it looks better and better all the time

[whtchocla7e]

GPU mining needs to die away slowly. Bring on the ASICs..

[Yanakitu Tenatako]

ASIC machines did not contribute anything, just taking $$$ out of minerts and put them in BFL-AVALON-KNC like companies.

Introducing ASIC to the scene would be same like cheating in the games, you will reach better score easily, but the game would become non-playable for a lot of ppl. Like when there is a cheater in CounterStrike pool and there are no admins, soon will be plenty of them, and the charm of gaming is spoiled by retard.

[yochdog]

I do have to say, ASICs indeed shift the profit from those doing the actual mining to those making the mining gear.

It is a bummer.  There is a lot of fun in designing, deploying, and maintaining a large GPU farm.  I will be sad when ASICs obliterate that niche.