Ethereum GPU mining?

Page 203 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Headfoot

Diamond Member
Feb 28, 2008
4,444
641
126
all I know is if I was in venezuela I would have 100% of my money in gold or crypto. I think that fundamentally is the value - distributed systems that can't be banned in reality even if governments try
 

ultimatebob

Lifer
Jul 1, 2001
25,135
2,445
126
Few tokens that are mineable will go to zero (ETH, ZEC, XMR). Maybe some of the fringe stuff like ECN is at risk? Some folks are obviously trying to spark a rally, but I see lower lows before we go back to the highs again.

I've seen it happen. I have a few thousand Nimiq (NIM), and it's value has dropped to the point where I couldn't cover the exchange fees to get the currency out of the exchange after I sold it all. So... it's basically worthless.
 

[DHT]Osiris

Lifer
Dec 15, 2015
14,004
12,071
146
Did any of you miners make any money mining after electric costs?
Yes, but I started about a year and a half-2 years ago. Paid off a 1080, decided it was probably worth pulling the trigger, and built a bigger rig. Paid that off about twice over, and trying to part it out now (as I don't need a pile of 1080ti's). That 'paid out twice over' is net, includes electricity and taxes.

Pity it's over, but it was a fun little ride.
 

ultimatebob

Lifer
Jul 1, 2001
25,135
2,445
126
Do you mean the three will not go to zero?



Chart says it wasn't very high to begin with.

It started trading at a little over half a penny each, now it's down to about 1/8th of a cent. I started mining it when it ICO'ed, hoping that it would go to a dollar. Wishful thinking :)

Initially, it didn't even have a price, since it wasn't traded anywhere. I knew that it was a gamble, but... damn.
 
Last edited:

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
It started trading at a little over half a penny each, now it's down to about 1/8th of a cent. I started mining it when it ICO'ed, hoping that it would go to a dollar. Wishful thinking :)

Penny stocks, including cryptos are a gamble. The only worst thing you can do is by using a slot machine. They are valued that low for a reason. I bought $100 worth of Penny stocks. Went down to $4, which was below the trading fee. Fortunately they said I won't be charged, so I was left with $4.

(Penny stocks are such because they didn't meet the listing requirements and got delisted from the big exchanges. Such delisting causes value to plummet even more. Chances of being even with your original investment, nevermind making money are very slim.)
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Eth was at $260 not long ago. Now its at $288. I think it'll stay stable, and start rising slowly before Constantinopole hard fork, then a proper rally will begin.

There's still a cloud hovering over the entire crypto market. The fear that over $2 billion in false market liquidity could leave at any instant is preventing much money from entering the market. Which is kind of interesting since it'll help separate projects from speculators, to an extent. It might actually be healthy for ETH to stay low for awhile until it matures into something more substantial than what we have today.

I've seen it happen. I have a few thousand Nimiq (NIM), and it's value has dropped to the point where I couldn't cover the exchange fees to get the currency out of the exchange after I sold it all. So... it's basically worthless.

Nimiq was never a smart play. It looked like a gimmick token from the start, backed by the idea that it was miner-friendly and browser-based. So no hunting for funky software from some guy on a forum somewhere, that may or may not be mining some of your blocks for himself (or whatever).

Other than that, it's yet another attempt by someone to introduce an entirely new blockchain ecosystem to tackle one or more problems. We already have ETH, NEO, EOS, XEM, and others. Do they really have any core concepts that make their blockchain fundamentally better than the others in any way? At a cursory glance, I would have to say no. BUT at least they have cash left over from their ICO, so even if NIM goes to $0, they can still continue development, if they wish.

Do you mean the three will not go to zero?

Yes. I do no think ETH, XMR, or ZEC will go to $0. I've had my doubts about ZEC, but it does endure.

Of all cryptos, XMR is the most likely to be banned in the US, though. And China, and possibly the EU . . .
 

ozzy702

Golden Member
Nov 1, 2011
1,151
530
136
Ethereum needs scaling and proof of stake asap. That said, with the negativity and FUD driven bear market, it's probably wise for the Ethereum devs to hold back good news/progress until things have calmed down a bit. Releasing during a bull trend makes a whole lot more sense and I wouldn't be surprised to see ETH gobble up marketshare if/when big upgrades are implemented.
 

Red Squirrel

No Lifer
May 24, 2003
67,198
12,027
126
www.anyf.ca
Penny stocks, including cryptos are a gamble. The only worst thing you can do is by using a slot machine. They are valued that low for a reason. I bought $100 worth of Penny stocks. Went down to $4, which was below the trading fee. Fortunately they said I won't be charged, so I was left with $4.

(Penny stocks are such because they didn't meet the listing requirements and got delisted from the big exchanges. Such delisting causes value to plummet even more. Chances of being even with your original investment, nevermind making money are very slim.)

Arn't a lot of penny stocks just because they are too new? Most stocks at one point probably started off that way. Ex: stuff on the TSX Venture. Too small to be on the real TSX but eventually think they get moved up? Honestly not sure though. I have like $500 in penny stocks in a local mining company. They started drilling a while back, have not heard much since then though. :eek: Figure it's better than spending that money at a fancy restaurant for a 1 day thing or something. :p
 

ultimatebob

Lifer
Jul 1, 2001
25,135
2,445
126
Arn't a lot of penny stocks just because they are too new? Most stocks at one point probably started off that way. Ex: stuff on the TSX Venture. Too small to be on the real TSX but eventually think they get moved up? Honestly not sure though. I have like $500 in penny stocks in a local mining company. They started drilling a while back, have not heard much since then though. :eek: Figure it's better than spending that money at a fancy restaurant for a 1 day thing or something. :p

Not really... most stocks IPO around the $20 a share range. If it's gone down to pennies per share, either something has seriously gone wrong with the business or they have some legal issues.

What I did was more like angel investing.... I put money into a venture before it was pre-revenue. Super high risk, but super high reward if it works out.
 

Red Squirrel

No Lifer
May 24, 2003
67,198
12,027
126
www.anyf.ca
Not really... most stocks IPO around the $20 a share range. If it's gone down to pennies per share, either something has seriously gone wrong with the business or they have some legal issues.

What I did was more like angel investing.... I put money into a venture before it was pre-revenue. Super high risk, but super high reward if it works out.

Oh I see. But yeah I realize what I'm doing is risky as well, figure I could easily just blow that money at a casino or something, at least with penny stocks I just sit on it and hope something happens.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Red Squirrel,

Yes, you can enter the company into such stock markets, but they are more or less obscure. If you want to list your company at a reputable market like Nasdaq(or TSX) for example, they have a requirement. Nasdaq, will delist your company if the share price stays under $1 for 30 days. Then you lose the spotlight which will cause a drastic plummeting of the share value as it disappears from the mass investing community's eyes. Of course, the fact that the company allowed itself to get to that state is the main reason why it drops so much.

So I've seen companies do reverse share split, to make it above $1, but ultimately the best way not to get delisted is having a proper business.

You'd want to do research to find out if such low valuation companies have such a history. Once they get delisted, the prior record may be challenging to find. It's up to you to find out whether its a real startup or it was a delisted one.

DrMrLordX,

It sounded like you were saying only 3 will go zero, but I figured that's not what you were saying.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
It sounded like you were saying only 3 will go zero, but I figured that's not what you were saying.

Oops, sorry for the confusion.

If anyone's interested, there are some sites that maintain lists of "dead" (or considered to be dead) crypto projects. The main ones are Coinopsy and DeadCoins. Maybe you can draw parallels between projects currently considered to be "dead" and ones that are still kicking around to sort out for yourself what will or will not fail.

I've observed that most "dead" projects either die because they were at least somewhat scammy (devs abscond with cash after ICO, more or less) or lack of developer interest. It may be that there are some projects on there classified as "dead" that simply have no presence on any exchanges and/or have practically nobody interested in their blockchain (where applicable). The dev(s) might still be making updates to the codebase anyway.

I would like to see a lot of the "dead" projects brought back into the world of the living, perhaps as a hideous mishmash of conflicting crypto concepts. Lichechain would be born! Or unborn as the case may be.

Hmm, one master blockchain to rule zombie subchains, with oracles and/or atomic transfers galore! A zombie fork of the Ethereum chain could contain all the dead ERC20 tokens out there. There're a LOT of those.
 

Red Squirrel

No Lifer
May 24, 2003
67,198
12,027
126
www.anyf.ca
Actually what happens if a crypto completely dies, like everyone stops mining. Could one person technically singlehandedly take it over? Like the whole idea behind how they work is that everything is validated by the network but if the network is more or less gone, one person can basically just do whatever they want with the data such as change balances. (probably require lot of computations for that). I guess at that point it's moot since if it died then nobody cares about it anymore.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Leaks are showing information about the next Ethash ASIC, the Innosilicon A10. The main chip is using Samsung's 10nm and presumably their own design as well. Price is said to be $5000 USD. Its said to do 485MH/s at ~850W of power consumption.

My verdict: This does change things from the Antminer E3, which no sane person should consider buying. At the efficiency they are suggesting, at the moment it seems viable. However, nothing much changes long term. Ignore the people saying algorithm should change to kill ASICs.

Reasons are numerous. One, because there's no guarantee the "ASIC" portion isn't more flexible then it sounds like, and that may mean minimal change may be needed from the manufacturer's perspective to make it run a new Ethash algorithm. The Ethereum team would waste precious time away from Casper and Sharding for at best, a short-term solution.

The biggest problem for Ethereum is scalability, and moving to Proof of Stake. Proof of Stake makes Level 1 scalability solutions like Sharding much easier and safer, thus they are complementary. Proof of Stake is the real anti-ASIC.

What about concerns from miners?
That their profits will be low and GPU mining eliminated? Not really relevant. Profits are already low, and the hard fork coming in less than 2 months(Constantinopole) likely has an issuance reduction, which will cut revenue more than Eth ASICs will. Let's see, an 84MH/s setup makes a puny $1.7 worth of ethereum per day! Before power costs! Issuance reduction by 1 Eth will reduce that by 1/3rd, and 2 Eth will reduce that by 2/3rd. Most will have to leave anyway. If most mining isn't done by large-scale companies already, any issuance reduction will render anyone except the most profitable large-scale facilities obsolete.

At $5000, it takes a 485MH/s device an incredible 700 days to reach ROI, and that's assuming difficulty stays same, and issuance stays the same at 3 Eth per block. Everyone, except the company making the Innosilicon A10 loses.

Possible discounts for the A10?

I have to assume the cost of production isn't that low. 485MH/s requires 5TB/s of memory bandwidth. GDDR5 is quite unlikely, as it requires 1W for every 10GB/s BW. That's 500W for the memory alone. Considering the E3 probably uses only 200W or so for memory, and more than 500W for the chip, I'm doubting it uses GDDR5, but rather HBM2. That would explain the exorbitant cost.

Next gen GPU threat.

A long term stable AMD RX 4xx/5xx card does 29MH/s. GTX 1060 using good memory gets 22-24MH/s. A GTX 2060 with GDDR6 memory at 14GT/s will likely get 38-40MH/s(maybe not right at launch as Nvidia card memory timings are locked and needs special software like OhGodAnETHlargementPill).

RTX 2070/2080 - 50+MH/s, RTX 2080 Ti - 80MH/s

RTX 2080 Ti is probably the most efficient and a 6-card system may do 480MH/s using 1200W. Not too far from the custom-silicon, maybe-algorithm-limited Innosilicon A10.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Actually what happens if a crypto completely dies, like everyone stops mining. Could one person technically singlehandedly take it over? Like the whole idea behind how they work is that everything is validated by the network but if the network is more or less gone, one person can basically just do whatever they want with the data such as change balances. (probably require lot of computations for that). I guess at that point it's moot since if it died then nobody cares about it anymore.

Depends on the underlying blockchain. But if you're looking at a "normal" blockchain, such as something that is basically a knockoff of Bitcoin (lots of those), then if there are no full nodes left, the entire blockchain history essentially goes offline. You have to have nodes running for the blockchain to exist and for the recording of validated transactions. Plus you have to have validators on top of that. But as long as you have at least one node running, the blockchain still lives.

Well... not every ERC20 token can be as awesome as say... Coolcoin :)

Not many projects can achieve that level of awesomenessness. One might argue that Coolcoin was never alive in the first place, though.

Unless you like t-shirts.

My verdict: This does change things from the Antminer E3, which no sane person should consider buying. At the efficiency they are suggesting, at the moment it seems viable. However, nothing much changes long term. Ignore the people saying algorithm should change to kill ASICs.

The EF wouldn't commit much time to it anyway. PoW is not in Ethereum's future.

Reasons are numerous. One, because there's no guarantee the "ASIC" portion isn't more flexible then it sounds like, and that may mean minimal change may be needed from the manufacturer's perspective to make it run a new Ethash algorithm. The Ethereum team would waste precious time away from Casper and Sharding for at best, a short-term solution.

If it truly is an ASIC, it isn't reprogrammable at all. But see above anyway.
Next gen GPU threat.


Interestingly enough, Nvidia has already warned investors not to expect sales to miners to reach the same heights that we saw in 2017 or 2018 (and those sales are already petering out).
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
The EF wouldn't commit much time to it anyway. PoW is not in Ethereum's future.

Yes, Vitalik and others are level minded, fortunately. I just wish intense public pressure won't cause them to buckle.

If it truly is an ASIC, it isn't reprogrammable at all. But see above anyway.

That's true. But I was thinking since its aiming at the Eth algorithm which is very memory bound, which is where it gets ASIC-resistance from, perhaps the designers might add some flexibility. Maybe an FPGA rather than ASIC. ASIC seems to be used as an umbrella term for anything more specialized than GPU in the press.

Interestingly enough, Nvidia has already warned investors not to expect sales to miners to reach the same heights that we saw in 2017 or 2018 (and those sales are already petering out).

Yup. Whatever is left is taking a vanishing share of profits.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
New Etherium fork? Or did I miss some nuance?

Coinbase now has a buy/sell, for BOTH "Etherium" and "Etherium Classic".

What's the difference?
 

Headfoot

Diamond Member
Feb 28, 2008
4,444
641
126
I think we will see crypto pop back up in a big way as soon as one major institutional investor decides to pour serious cash in it. That alone I think would be enough to trigger a round of FOMO across amateurs and professionals alike

Can you imagine what would happen if the Saudi or Norweigan soverign wealth fund said they were buying some multi million stake in a basket of cryptos?
 

Feld

Senior member
Aug 6, 2015
287
95
101
New Etherium fork? Or did I miss some nuance?

Coinbase now has a buy/sell, for BOTH "Etherium" and "Etherium Classic".

What's the difference?
Ethereum is the real one. The "Classic" one isn't new - it's the old version of the chain from before the fork that recovered the stolen DAO funds in 2016. Coinbase just decided to add it recently because it was technically easy for them to do. That old fork of the chain was supposed to die off back then but a small minority has maintained running it anyway. It mostly exists as a means of scamming people and to serve as a haven for "code is law" ultra purists who either don't mind theft or don't believe theft is even a legitimate concept, and are philosophically unwilling to fix the results of even an obvious error in code. The chain has almost no developers or dapps being developed for it, so it is never going to amount to anything. But it is crypto, so it's possible to make money speculating on it if that's your bag.
 
  • Like
Reactions: harobikes333

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
That's true. But I was thinking since its aiming at the Eth algorithm which is very memory bound, which is where it gets ASIC-resistance from, perhaps the designers might add some flexibility. Maybe an FPGA rather than ASIC. ASIC seems to be used as an umbrella term for anything more specialized than GPU in the press.

I wouldn't expect a full FPGA, but something closer to a mobile processor (lots of fixed-function hardware grafted to a general-purpose CPU). Remember that an FPGA is actually the most flexible computing mechanism you can buy, which is why they are so expensive to implement and require so many transistors. You can effectively model any kind of CPU/GPU you want with them, given the time and talent. ASICs are on the other end of the spectrum: fix-function hardware designed to run a tiny set of instructions repeatedly, at (generally) high speeds. That's what makes them so cheap versus GPUs and general-purpose CPUs: no need for fancy cache systems, brand predictors, etc. You know what you're going to run in advance.

With Ethereum, most of the expense of building hardware to handle its current algorithm involves memory controllers and memory. You want low latency and high bandwidth.

I think we will see crypto pop back up in a big way as soon as one major institutional investor decides to pour serious cash in it. That alone I think would be enough to trigger a round of FOMO across amateurs and professionals alike

Can you imagine what would happen if the Saudi or Norweigan soverign wealth fund said they were buying some multi million stake in a basket of cryptos?

Remember that institutional investors like to buy low just as much as the rest of us. They're not going to buy the highs like idiots. They saw how cheap crypto was in 2016, and I strongly suspect that they want to see those prices again before they make any significant capital commitments.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
With Ethereum, most of the expense of building hardware to handle its current algorithm involves memory controllers and memory. You want low latency and high bandwidth.

Just saying, assuming it can only run one algorithm, and literally nothing else if the devs decide to hard fork away from it isn't looking at reality. That the chip or the system may be relatively easy for the manufacturer to make it run on the new algorithm. They might introduce a version 1.01 version with slight firmware adjustments 6 months later and the devs would have totally wasted time on it.

That is actually what the devs have said, and why they want to avoid changing the algorithm.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Just saying, assuming it can only run one algorithm, and literally nothing else if the devs decide to hard fork away from it isn't looking at reality. That the chip or the system may be relatively easy for the manufacturer to make it run on the new algorithm. They might introduce a version 1.01 version with slight firmware adjustments 6 months later and the devs would have totally wasted time on it.

That is actually what the devs have said, and why they want to avoid changing the algorithm.

It's possible, and there are a number of different approaches they could take to achieve that goal. One would be to strip down a GPU and "guess" at which transistors wouldn't be needed for cryptographic number crunching. Another would be to model a number of ASICs and tie them in to a small general-purpose CPU. Another would be just to strap a fast memory controller to a general purpose CPU and see how fast they could get it to crunch numbers without much overall throughput. You get the idea.

That being said, when Monero wiped out the Bitmain XMR ASICs with an algorithm chain, Bitmain did not strike back, so obviously they weren't prepared there either.

Monero network hashrate did actually fell by 80% from 1GH/s to 200MH/s after april anti-Bitmain hardfork.

But the opportunity was almost immediatelly picked by miner and multi-pools so it went to 400MH/s.

Since april 2018, I gave up on ETH mining and went back to Monero.
With http://www.xmrcgpu.com/XMRCGPU.zip I earn lot more on GPU's and I earn quite bit with CPU setting affininty to CPU 0,2,4,6,8,10 (Miner eats cache of cores, but I have hyperthreading "threads" left alone, way enough to browse the web or watch movies).

I also run XMRcgpu on unused cores on servers I manage and that's really awesome.

Glad you brought that up. BTW if you want to bring any XMR mining news to the table, we do have a somewhat-decrepit thread for that:

https://forums.anandtech.com/threads/xmr-mining-thread.2520539/