Ethereum GPU mining?

Page 213 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Apr 27, 2000
12,759
1,563
126
Yeah, right now depending on your cost of electricity it's around $140-$200 per ETH to mine based on just electricity, not even factoring hardware costs. ETH will need to get around ~ $750 before we see another mining craze.
At this point, it's cheaper to mine with ASICs. dGPUs should be pointed at something else.
 

ozzy702

Senior member
Nov 1, 2011
986
203
136
At this point, it's cheaper to mine with ASICs. dGPUs should be pointed at something else.

Yup. Currently way more profit to be had with other algos, but I don't think that precludes another mining craze centered around ETH. It all just depends upon price. If ETH shoots towards the $1,000 range we'll see the same rush to pickup hardware even if the profits aren't what we enjoyed in 2016/2017.
 
Apr 27, 2000
12,759
1,563
126
If ETH shoots towards the $1,000 range we'll see the same rush to pickup hardware even if the profits aren't what we enjoyed in 2016/2017.
Yeah, but which hardware? I wouldn't get anything but an ASIC even at those price levels.
 

ozzy702

Senior member
Nov 1, 2011
986
203
136
Yeah, but which hardware? I wouldn't get anything but an ASIC even at those price levels.
Depends on what's available. Historically ASICs have been hard to come by when demand spikes. 7nm NAVI could be decent depending upon price and if ASICs aren't available I could see another run on GPUs. Again, it's all dependent upon how high crypto climbs.
 
Apr 24, 2019
121
129
76
At this point, it's cheaper to mine with ASICs. dGPUs should be pointed at something else.
I'm not doubting you, but everywhere I look has ASICs losing money even with my relatively cheap $0.09/kWH electricity cost.

Further, unless you're the first to receive shipment of ASIC, ROI will be tanked once everyone is running the same ASIC and algo because difficulty will scale up.

Also, the ASIC isn't repurposable, so if the coin you're mining/algo fails, you're done and out hundreds or thousands of dollars.

FPGAs hold hope, but are a ways off.

dGPUs still have a role, as long as you're buying them for a separate purpose, IMO. They're reprogrammable, can game, and fit in your PC already.
 
Apr 27, 2000
12,759
1,563
126
I'm not doubting you, but everywhere I look has ASICs losing money even with my relatively cheap $0.09/kWH electricity cost.
At current prices? I'm not surprised. And are you looking at ETH ASICs or BTC ASICs? The latest ETH ASICs are much more efficient at mining than any dGPU on the market, so if the ROI is that bad on the ETH ASICs, it's gonna be worse on a dGPU.

Further, unless you're the first to receive shipment of ASIC, ROI will be tanked once everyone is running the same ASIC and algo because difficulty will scale up.
Only true if price stays low. Beyond a certain price, mining becomes profitable. But dGPUs will fare worse in this equation.

Also, the ASIC isn't repurposable, so if the coin you're mining/algo fails, you're done and out hundreds or thousands of dollars.
True, but the problem here is that, at the present time, ETH has already failed for dGPUs (for the most part). You have to make them mine something else from the get-go. Mining ETH on dGPUs is not presently viable, and won't be until well after ETH ASICs have started paying off.

Eventually ETH will go to some PoW variant, making all mining non-existent.

dGPUs still have a role, as long as you're buying them for a separate purpose, IMO.
Most people who mine are not getting dGPUs for a separate purpose. I know when I had 11 cards mining at once, that I only ever used one of them for actual display duties. Perhaps that is something that is wrong with PoW: people pile up so much mining hardware. There's no check or balance to prevent people from doing such a thing. Not even sure how that would work. It would be interesting if people could only mine with one card or one CPU.
 

Accord99

Platinum Member
Jul 2, 2001
2,167
11
91
The Innosilicon A10's paper specs is more efficient than any video card, but despite being announced in the summer last year, it seems to have never shown up, at least to the public. The other miner, the Antminer E3 (200 Mhash/800W) gets beaten by 4x1080 Ti in performance and performance/watt, and be completely crushed by the Radeon VII.
 
Apr 27, 2000
12,759
1,563
126
The Innosilicon A10's paper specs is more efficient than any video card, but despite being announced in the summer last year, it seems to have never shown up, at least to the public.
Probably sold out to select buyers, if it was ever produced. Hey at least Radeon VII is widely available now, yay! Happy mining people.
 
Aug 25, 2001
44,463
807
126
Perhaps that is something that is wrong with PoW: people pile up so much mining hardware. There's no check or balance to prevent people from doing such a thing. Not even sure how that would work. It would be interesting if people could only mine with one card or one CPU.
Dah, Comrade! Let's limit people's paychecks, or limit their work hours, or how much stock they can own.





/s
 
Oct 27, 2006
19,794
302
126
I hope GPU mining stays largely dormant if possible. Or at least instead of doing largely junk math (other than blockchain maintenance), have it crunching medical research, protein folding, SETI, just anything not literally just wasting power.
 
Apr 27, 2000
12,759
1,563
126
Dah, Comrade! Let's limit people's paychecks, or limit their work hours, or how much stock they can own.
/s
Da, Tovarishch. You are ready for job in politburo.

Seriously though, forcibly distributing verification like that makes 51% attacks less likely. Look at what happened to BTC and BCH with their stupid ASIC nonsense. It's horrible.

I hope GPU mining stays largely dormant if possible. Or at least instead of doing largely junk math (other than blockchain maintenance), have it crunching medical research, protein folding, SETI, just anything not literally just wasting power.
CureCoin tried that, but it never took off.
 
Aug 25, 2001
44,463
807
126
Seriously though, forcibly distributing verification like that makes 51% attacks less likely. Look at what happened to BTC and BCH with their stupid ASIC nonsense. It's horrible.
Oh, I see what you're saying now. I thought that you went off on some anti-American rant, very uncharacteristic of you.

Maybe some of the recent experiments in "alternative voting schemes" research might play a part here?
 
Apr 27, 2000
12,759
1,563
126
Oh, I see what you're saying now. I thought that you went off on some anti-American rant, very uncharacteristic of you.
Between 51% attacks and the overall wastefulness of PoW, it has some highly-undesirable characteristics regardless of your national or political allegiances. It isn't sustainable behavior for any digital currency system (even one as slow as BTC).

Maybe some of the recent experiments in "alternative voting schemes" research might play a part here?
Possibly. PoS obviously has a lot of interest, but it has problems of its own. And obviously those problems are great enough that the Ethereum Foundation hasn't rolled out ANY of their PoS schemes to the mainnet yet.
 
Oct 27, 2006
19,794
302
126
:)

I mean, something left unsaid during the mega GPU mining days of early BTC and then the recent ETH insanity, we're talking tens of millions of GPUs grinding away. Probably easily the most powerful network in existence

Seems mind bogglingly wasteful to use that unprecedented power to ... mostly run in circles with no scientific purpose. Hell, what about a mining coin that used spare computer power to lease for rendering projects for students or hire it out for pro use. IDK, spitballing. I know the minor altcoins that have tried it, but nothing major :/

Ah, found d an article from 2013 showing that BTC Crush was already at exascale, being more powerful than the top 500 global supercomputers combined. In fact, all 500 of them only represented just under 12% of the BTC power.

Now that was WAY back in 2013 lol. It wouldn't surprise me one bit if there was 1000x+ more power by fall 2017 in GPU mining, probably much higher. Man, imagine if you harnessed that for deep learning or work towards AGI.
 
Last edited:
Apr 27, 2000
12,759
1,563
126
The problem with a lot of BTC miners is that they're nothing but SHA256 ASICs. There's limits to what you can do with those.
 
Oct 14, 2003
6,257
357
126
Seems mind bogglingly wasteful to use that unprecedented power to ... mostly run in circles with no scientific purpose.
Whether its truly wasteful depends on how you view the value of sustaining existing Crypto-based networks. If the networks are worth it, then the spending is justified.

have it crunching medical research, protein folding, SETI, just anything not literally just wasting power.
Setting up a crypto mining farm is far simpler than a GPU farm for scientific work. The differences are so enormous that they are not comparable.

Setting up a GPU rig to mine crypto just requires one program, with few command lines. You run a single algorithm that's much much easier to run it at peak, and most of the job is done by the software developer that makes the miner.

To allow a public GPU sharing farm to happen on a mass scale as crypto mining does, not only it needs to be easy both for sellers and buyers, you need the buyers to be incentivized in a way comparable to crypto mining. The scale would also be big enough that huge datacenters(you know the ones that go in Top 500 for example) would start to be displaced. That's not happening anytime soon, maybe never even.

Golem, an ERC based token for such ideas, has much more sellers than buyers.
 

coercitiv

Diamond Member
Jan 24, 2014
3,359
887
136
we're talking tens of millions of GPUs grinding away. Probably easily the most powerful network in existence

Seems mind bogglingly wasteful to use that unprecedented power to ... mostly run in circles with no scientific purpose.
What I find really insulting in this theory that mining crypto coins is just a waste of energy is the actual purpose of the hardware ---> exchanging energy for fun. So if I play games in my free time I'm a gamer who spends his/her money to indirectly advance silicon based tech, but if I'm a miner in my free time... I'm wasting resources.

People who criticize crypto mining energy usage should take a step back and analyze what entertainment is all about around the world. The amount of energy we waste every day to justify our existence is incredible. We waste precious water, we waste energy, we waste lives every day because life is hard and such is the human condition. We also understand that from a macro level, the smart thing to do is not to force people into behavioral templates (fails every freaking time), but rather change the system to maintain a healthy status quo. Want less waste of energy? Use whatever power you have (pun intended) to switch global energy production to green, renewable sources only. Energy cost will change towards the real cost, entertainment energy cost will follow, and crypto coin energy usage will adjust accordingly. Energy conservation at it's best.
 
Aug 25, 2001
44,463
807
126
Yeah, how many people will go online, cry about the "Waste of energy" from mining, and then go on a scenic drive in the Fall?
 
Apr 27, 2000
12,759
1,563
126
What I find really insulting in this theory that mining crypto coins is just a waste of energy is the actual purpose of the hardware ---> exchanging energy for fun. So if I play games in my free time I'm a gamer who spends his/her money to indirectly advance silicon based tech, but if I'm a miner in my free time... I'm wasting resources.
In the case of BTC at least, they had people in areas with industrial hydro contracts going online to mine in old factories and stuff. That's mass-scale energy consumption. Not everyone was overly-pleased with that.

Yeah we do burn a lot of energy on frivolous stuff like . . . movies or video games. Also, nobody has said how much power it takes to run existing currency exchange technology, like SWIFT or trucking around paper currency.
 
Oct 27, 2006
19,794
302
126
In the case of BTC at least, they had people in areas with industrial hydro contracts going online to mine in old factories and stuff. That's mass-scale energy consumption. Not everyone was overly-pleased with that.

Yeah we do burn a lot of energy on frivolous stuff like . . . movies or video games. Also, nobody has said how much power it takes to run existing currency exchange technology, like SWIFT or trucking around paper currency.
Exactly. And if you look at my comments, it's not saying people shouldn't be allowed to mine, but rather that it's a shame that SO very much compute power is going largely to junk math.

It's not even close to a 1:1 for gaming either. Mining GPU rig setups can vary from one to dozens for a home user, to hundreds of thousands with the larger outfits, running 100% 24/7/365.

I have an 8086k/1080ti, and 2700x/Vega56 pair of rigs. I game if I'm lucky 1-2hrs a week, but more often 1-2hrs a month.

Compare that to 168hrs/week, or 670+hrs/month for mining.

Now I do believe the values in blockchain will more effectively harness this type of resource in the future. At least that's my sincere hope.

One can be a fan of crypto and mining while still admitting it's currently pretty wasteful on the energy side of things, and without the need for strawmanning or false equivalency.


Wired
 
Oct 14, 2003
6,257
357
126
One can be a fan of crypto and mining while still admitting it's currently pretty wasteful on the energy side of things, and without the need for strawmanning or false equivalency.
Yes, and if they thought about it, maybe Vitalik would have had PoS version on Ethereum's birthdate.

But the one who created Bitcoin is a pioneer, and Vitalik is one too. While hindsight tells us that it could have been done better, maybe thinking that way might have resulted in nothing done in the first place. You crawl before you walk. Once you satisfied the need for oxygen, then you want water. Then food, shelter, etc. Thinking about the environmental effects are really the higher level thought that comes after lower level ones like creating the thing are fulfilled.

Progress is always gradual. In the point of view of the observer that pays little attention, it may look like a giant leap out of nowhere. But its not.
 


ASK THE COMMUNITY

TRENDING THREADS