Ethereum GPU mining?

[Madpacket]

Because he would probably also need an new PSU and maybe even mobo. he already has a card so he needs 3 Pcie 16x slots (at least 16x size, not speed).

Mining is fine to pay off a card you own anyway. Else if you want to make bank, just invest some money ($1k+) and wait.

No he would just need a few powered 1X-16X risers. They work fine in most 1X PCIe slots. Sometimes you need to make a "dummy plug" to get Windows to recognize the cards but you can make or buy these for cheap. I agree if he needs a new power supply it's not really worth it. A good 850W power supply with at least 75 amps on the +12 should easily handle two 390's comfortably but three could be pushing it even with undervolting. Topping out a 850W with a 4GB 380 may be a smarter buy as it saves from having to buy a 1KW supply, and 380's use a lot less energy under loads. An extra 20Mh would net a total of around 80Mh machine for 850W. Not bad.

If you're serious and really believe in the future of Ethereum the best approach is both mining and buying IMO. The mining part is dependent on many factors (cost of electricity, cost of hardware, warm or cold climate, noise pollution etc.) where the purchase of Ethereum is a much more straight forward endeavor. However all of our situations are unique so you need to assess your own factors and decide.

Mining is still very much worth it if the above factors play in your favour but there's always a time limit on mining profitability due to a number of factors which have been discussed Ad nauseam and not worth repeating.

I personally bought as much Ether as I plan on mining in total (so around 50/50 split). If Ethereum tanks in price your investment in simply buying Ethereum may lose you more money compared to buying a few mining machines which costs can be recouped from reselling hardware. On the flip side all efforts towards mining will seem silly compared to buying (at today's price) if the price of Ethereum shoots up to 50 USD a coin.

 

[Madpacket]

A tip for the people who live in expensive hydro areas..

Do you have any trusting friends in need of a new video card? Do they have a computer with a decent power supply and only casually game? If so sell them a mining card for half the price they normally sell for but with the understanding they have to keep their computers on 24/7 for at least 3-4 moths. This way you offset the cost of the card by recouping half of it on point of sale and pay no electricity costs for the mining period agreed on which will help you recoup the remaining costs even faster.

Your friend is happy because he gets an awesome new card for half the price and only has to run a program for a while before he can turn it off (or he can start mining for himself if it's still profitable at the end of the agreed timing). Just make sure the cost of electricity does not exceed the second half of the card cost during the mining period or you won't be a very good friend. This also takes out the worry of having to resell the card down the road

 

[Madpacket]

To any moderator following along here...

This thread has 30,000 views, is 19 pages long and doesn't look like it's going to die anytime soon.

I'm not sure if the OP (thanks for posting here metalliax!) knew this would happen but do you think its time we get our own section or subsection on cryptocurrency or cryptomining?

Sure a lot of what we're talking about is video card related (as that is the current vessel for generating coins) but the zeitgeist of the thread is more about optimizing mining, mining pools, mining tweaks, tips and tricks, Ethereum speculation etc.

Just a suggestion.

 

[Hugo Drax]

sounds like a lot of work for 150 bucks a month.

 

[DrMrLordX]

thanks.

Ill have to grab something to monitor the connection. Im used to cgminer and its stratum connection using only a few kb/s bandwidth on btc mining, so this is way different.

Ill setup a stratum proxy at some point and see what happens. new to eth mining.. mining it mainly to offset the price of the r9 390

Let us know your results and if you manage to find a way to cut back on bandwidth usage. You can't be the only one out there under some kind of bandwidth constraints.

I am using about 575w at the wall. That is with 4 drives, 6 fans, 1 water pump, and the system listed in my sig, so I'm guessing I'm pulling about 500w for the two cards at the wall.
If I go -50mV (at stock 947/1250) I use about 100w less total.

That's not too bad actually. 390s at -100 mv and 1100/1500 chew up 240W DC and more AC at the wall.

On the flip side all efforts towards mining will seem silly compared to buying (at today's price) if the price of Ethereum shoots up to 50 USD a coin.

When it was $1 per ether, buying was 100% the way to go. Now it seems like mining might be the better investment, though if it keeps going up, buying will prove to have been the best choice even now.

sounds like a lot of work for 150 bucks a month.

It isn't really, especially if you're having fun doing it. Once things are up and running, it's just fire and forget, until something breaks . . .

 

[MrTeal]

No he would just need a few powered 1X-16X risers. They work fine in most 1X PCIe slots. Sometimes you need to make a "dummy plug" to get Windows to recognize the cards but you can make or buy these for cheap. I agree if he needs a new power supply it's not really worth it. A good 850W power supply with at least 75 amps on the +12 should easily handle two 390's comfortably but three could be pushing it even with undervolting. Topping out a 850W with a 4GB 380 may be a smarter buy as it saves from having to buy a 1KW supply, and 380's use a lot less energy under loads. An extra 20Mh would net a total of around 80Mh machine for 850W. Not bad.

If you're serious and really believe in the future of Ethereum the best approach is both mining and buying IMO. The mining part is dependent on many factors (cost of electricity, cost of hardware, warm or cold climate, noise pollution etc.) where the purchase of Ethereum is a much more straight forward endeavor. However all of our situations are unique so you need to assess your own factors and decide.

Mining is still very much worth it if the above factors play in your favour but there's always a time limit on mining profitability due to a number of factors which have been discussed Ad nauseam and not worth repeating.

I personally bought as much Ether as I plan on mining in total (so around 50/50 split). If Ethereum tanks in price your investment in simply buying Ethereum may lose you more money compared to buying a few mining machines which costs can be recouped from reselling hardware. On the flip side all efforts towards mining will seem silly compared to buying (at today's price) if the price of Ethereum shoots up to 50 USD a coin.

Powered risers aren't for everyone though. They force you to make certain compromises in the way you physically mount things, and it might make it impossible to keep everything enclosed in a standard ATX case. Really, if someone's considering a small foray into mining and doesn't want the mess of an open air system, with free electricity I'd consider buying/building a PC with case for $200-250 and tossing the two 290s into that. You could put together something like the following system cheap, and with two x16 slots triple spaced, it's just a matter of plugging the cards in, tossing on Ubuntu, and get mining. It's a little less messy that going the full route with suspended cards and risers.
CPU: Intel Pentium G4400 3.3GHz Dual-Core Processor ($59.99 @ SuperBiiz)
Motherboard: ASRock B150 PRO4/D3 ATX LGA1151 Motherboard ($74.98 @ Newegg)
Memory: A-Data XPG V1.0 8GB (2 x 4GB) DDR3-1600 Memory ($29.99 @ Amazon)
Storage: Western Digital AV-GP 250GB 3.5" 5400RPM Internal Hard Drive ($19.95 @ Amazon)
Case: Thermaltake VL80001W2Z ATX Mid Tower Case ($22.99 @ Micro Center)
Power Supply: EVGA 600B 600W 80+ Bronze Certified ATX Power Supply ($45.98 @ Newegg)
Total: $253.88
Prices include shipping, taxes, and discounts when available
Added benefit of this is that should you decide to get out, you can probably offload the whole system without taking a very large hit.

The battle between buying and mining really isn't much different than it was for BTC. You're essentially staking the money at the moment of purchase. If you buy a 380 at $180 and today 1 Eth is $11, you're betting that it will net 16.4Eth or more after power costs. It's definitely hedged though, since as you noted unlike BTC ASICs the hardware itself has a residual value even if Eth drops to pennies.

Even if Eth does jump to $50 though, that doesn't make buying mining hardware a bad choice vis-a-vis just buying Eth. It's not unreasonable to expect mining to produce as much Eth as buying for a given dollar, even if the price should go up. A $180 380 can be expected to earn 10Eth/month right now, and even if the daily earnings were to drop by half each month you'd still reasonably expect to earn ~13Eth or so in the next 3 months. Doubling each month might be a little pessimistic though; it's a pretty big ask outside the development of an ASIC. If in a month from now at the start of May the network has doubled again to 3500GH/s and the price jumps to $50, I question whether the network would be able to respond more than even the proposed doubling from 3500GH/s to 7000GH/s. That's already a jump equivalent to over 100,000 390s. That's a lot of additional hardware in a month, considering that Jon Peddie Research reported that just under 6M enthusiast class GPUs were sold in all of 2015.

 

[metalliax]

I tested my Fury Nano a bit more last night and put together a small table comparing hashrates at different power settings. I also took at-the-wall readings for both sytem idle (screen off, system on) and mining. Note that for the below table, I used 26W at my idle system power (I have a 6700k w/ a platinum rated power supply). I also used $11.31 as current ETH price and 25.38 as current Th difficulty. Obviously this is changing in real-time. And for power, I have table assuming 11 cents/kwh.

I was most surprised that the Fury Nano only draws ~91W while mining at almost 20 Mh/sec.



Edit: I originally stated my kwh cost @ 33 cents/kwh but realized I had set it to 11 cents/kwh in the graph.

Edit 2: I added a similar graph for my 290x.

 

[Erenhardt]

When it was $1 per ether, buying was 100% the way to go. Now it seems like mining might be the better investment, though if it keeps going up, buying will prove to have been the best choice even now.

Yes, but it is not how it works. I have $4000 in a pocket, If I knew ETH would go from $1 to $11 I would totally buy $4000 worth of ETH.

Alternatively, will eth go to $100 in next month? If so I'm all in.

No one knows really, that is why mining is a safe bet. You can only loose on the hardware value degradation and electricity if it drops.

 

[DrMrLordX]

Yes, but it is not how it works. I have $4000 in a pocket, If I knew ETH would go from $1 to $11 I would totally buy $4000 worth of ETH.

Alternatively, will eth go to $100 in next month? If so I'm all in.

No one knows really, that is why mining is a safe bet. You can only loose on the hardware value degradation and electricity if it drops.

True! Hindsight is 20/20. Nobody knows where eth will go from here. Probably up, but is that certain . . . ?

Also: great table there metalliax! That should help anyone who is seriously considering mining with a Nano.

For you Linux miners: have you found any software solutions for Hawaii that controls voltage on a card that allows it, such as the MSI 390? I see that you can change PowerTune levels, but I'm not sure that's anywhere near as useful as being able to change voltage.

 

[metalliax]

Also: great table there metalliax! That should help anyone who is seriously considering mining with a Nano.

No problem... I just edited my previous post and added a table for my 290x with the same baseline stats.

 

[EightySix Four]

sounds like a lot of work for 150 bucks a month.

At current prices I'm doing closer to $600 and spending $0 on electricity.

 

[VeryCharBroiled]

[bandwidth usage]

Let us know your results and if you manage to find a way to cut back on bandwidth usage. You can't be the only one out there under some kind of bandwidth constraints.

just started monitoring (using netbalancer free) and my single r9 390 at around 30mhs is 1 KB/s up and 1KB/s down.

BTW it was mentioned the DAG is downloaded.. Im not so sure.. it would of taken about an hour for me to dl 1 gig. my system created one in a few minutes.

still learning the ropes so please ignore any obvious gaffs.

 

[DrMrLordX]

No problem... I just edited my previous post and added a table for my 290x with the same baseline stats.

Nice! That'll come in handy for someone. Got any 290s?

[bandwidth usage]


just started monitoring (using netbalancer free) and my single r9 390 at around 30mhs is 1 KB/s up and 1KB/s down.

Hmm okay, that's 172.8 Mb / day in aggregate. That's actually less than what I'm pushing in aggregate assuming linear scaling of bandwidth requirements per MH/s, so . . . huh. Interesting.

BTW it was mentioned the DAG is downloaded.. Im not so sure.. it would of taken about an hour for me to dl 1 gig. my system created one in a few minutes.

still learning the ropes so please ignore any obvious gaffs.

Hey I'm learning too. Not sure what to make of the DAG . . . that was an assumption on my part. I wonder what geth.exe's traffic was from then?

 

[Accord99]

Hey I'm learning too. Not sure what to make of the DAG . . . that was an assumption on my part. I wonder what geth.exe's traffic was from then?

Was it catching up with the block chain?

 

[Elfear]

Can Nvidia and AMD cards mine in the same system?

 

[DrMrLordX]

Was it catching up with the block chain?

Possibly. I wasn't paying much attention to bandwidth usage until a day or two ago so I was not able to monitor usage as I was setting up everything.

Can Nvidia and AMD cards mine in the same system?

Yes. You must assign them to separate clients since they rely on different platforms.

 

[metalliax]

Nice! That'll come in handy for someone. Got any 290s?

My brother has 2x 390s mining. I can see if i can get some stats from his rig. I actually just sold my 290x as the warm season is about to hit me and it's already quite pricey to mine at 33c/kwh here in the bay area.

 

[Madpacket]

Powered risers aren't for everyone though. They force you to make certain compromises in the way you physically mount things, and it might make it impossible to keep everything enclosed in a standard ATX case. Really, if someone's considering a small foray into mining and doesn't want the mess of an open air system, with free electricity I'd consider buying/building a PC with case for $200-250 and tossing the two 290s into that. You could put together something like the following system cheap, and with two x16 slots triple spaced, it's just a matter of plugging the cards in, tossing on Ubuntu, and get mining. It's a little less messy that going the full route with suspended cards and risers.
CPU: Intel Pentium G4400 3.3GHz Dual-Core Processor ($59.99 @ SuperBiiz)
Motherboard: ASRock B150 PRO4/D3 ATX LGA1151 Motherboard ($74.98 @ Newegg)
Memory: A-Data XPG V1.0 8GB (2 x 4GB) DDR3-1600 Memory ($29.99 @ Amazon)
Storage: Western Digital AV-GP 250GB 3.5" 5400RPM Internal Hard Drive ($19.95 @ Amazon)
Case: Thermaltake VL80001W2Z ATX Mid Tower Case ($22.99 @ Micro Center)
Power Supply: EVGA 600B 600W 80+ Bronze Certified ATX Power Supply ($45.98 @ Newegg)
Total: $253.88
Prices include shipping, taxes, and discounts when available
Added benefit of this is that should you decide to get out, you can probably offload the whole system without taking a very large hit.

The battle between buying and mining really isn't much different than it was for BTC. You're essentially staking the money at the moment of purchase. If you buy a 380 at $180 and today 1 Eth is $11, you're betting that it will net 16.4Eth or more after power costs. It's definitely hedged though, since as you noted unlike BTC ASICs the hardware itself has a residual value even if Eth drops to pennies.

Even if Eth does jump to $50 though, that doesn't make buying mining hardware a bad choice vis-a-vis just buying Eth. It's not unreasonable to expect mining to produce as much Eth as buying for a given dollar, even if the price should go up. A $180 380 can be expected to earn 10Eth/month right now, and even if the daily earnings were to drop by half each month you'd still reasonably expect to earn ~13Eth or so in the next 3 months. Doubling each month might be a little pessimistic though; it's a pretty big ask outside the development of an ASIC. If in a month from now at the start of May the network has doubled again to 3500GH/s and the price jumps to $50, I question whether the network would be able to respond more than even the proposed doubling from 3500GH/s to 7000GH/s. That's already a jump equivalent to over 100,000 390s. That's a lot of additional hardware in a month, considering that Jon Peddie Research reported that just under 6M enthusiast class GPUs were sold in all of 2015.

Good analysis Mr.Teal. And I generally agree on most points with you where all my mining machines are in their own cases in PCIEx16 slots etc. However the adapters aren't that expensive and all you need is a milk crate, a Dremel like tool or file, a few zip ties and a bit of time to build a proper open air rig that can accommodate between 4-6 cards.

Here's a short tutorial video of a Litecoin rig. You can copy this method for Ethereum mining.

https://www.youtube.com/watch?v=fzK-twEBKF4

This makes the issue of fitting many cards in one case a moot point and all you really need is a $20.00 house fan to blow air across your rig for good cooling. It's probably the cheapest way to build a dedicated mining rig.

Re difficulty, there are a ton of older Radeon cards still out there that are perfectly capable of mining. These cards were released around late 2012 / early 2013 and sold for a long time so who knows how many will be used for Ethereum mining? The difficulty will continue to rise as long as it's profitable, at what rate? well that depends on the price of Eth. The biggest concern would be an ASIC release but I don't think we'll see the same situation as with Bitcoin or Litecoin as the Ethereum core devs are very keen on coming up with proof of stake which could change things pretty dramatically.

Fun times...

 

[MrTeal]

Good analysis Mr.Teal. And I generally agree on most points with you where all my mining machines are in their own cases in PCIEx16 slots etc. However the adapters aren't that expensive and all you need is a milk crate, a Dremel like tool or file, a few zip ties and a bit of time to build a proper open air rig that can accommodate between 4-6 cards.

Here's a short tutorial video of a Litecoin rig. You can copy this method for Ethereum mining.

https://www.youtube.com/watch?v=fzK-twEBKF4

This makes the issue of fitting many cards in one case a moot point and all you really need is a $20.00 house fan to blow air across your rig for good cooling. It's probably the cheapest way to build a dedicated mining rig.

Re difficulty, there are a ton of older Radeon cards still out there that are perfectly capable of mining. These cards were released around late 2012 / early 2013 and sold for a long time so who knows how many will be used for Ethereum mining? The difficulty will continue to rise as long as it's profitable, at what rate? well that depends on the price of Eth. The biggest concern would be an ASIC release but I don't think we'll see the same situation as with Bitcoin or Litecoin as the Ethereum core devs are very keen on coming up with proof of stake which could change things pretty dramatically.

Fun times...

Yeah, though the hardest part about milk crate mining can be finding a source of milk crates. If you're that kind of miner though, you already know it.

One thing I found about risers is that you have to account for the lead time in the costing of them, as well. There's a bunch of suppliers in China that will sell you them at reasonable prices, if you're willing to wait for a month to get them. In the US it's probably a little better, but for a Canadian choices for USB3 cabled risers is pretty limited from my searching. You can pick up a Z97 board with 3 physical x16 slots for $100 US or the even cheap Skylake board I listed above for $75, and get them to your door next day for a pretty reasonable cost. Assuming that the R9 290's are local and wouldn't be shipped slowly, now and mining tomorrow on $250 of new hardware might be a better value proposition than ordering $30 worth of powered risers, waiting a week, and integrating them into your existing system. A week of 60MH/s is US$77 right now, which takes a big chunk out of the purchase price of that hardware.
I bought a few more risers myself, but looking at the numbers and on Kijiji I found it better to just pick up some old hardware to get mining quickly rather than wait on risers to maximize the number of slots I use on a board. The downside is larger upfront cost and slightly worse power usage, as well as decreased density.


I wonder about the number of 7xxx cards out there that are capable of mining. It's obviously limited to Tahiti and some of the Pitcairn chips with at least 2GB of memory, but that's still a pretty large number. The supply of used cards has dried up considerably in the last couple months though. I have to imagine that at this point, almost every GPU mining rig is already pointed at Eth. There's going to be a really large number of cards that simply are not going to get switched over regardless of the price as they're in people's gaming rigs and they just don't care about crypto.
The nVidia market might be considerably larger though. A 680/770/960 might only net ~10MH/s while a 970/980 is good for closer to 20MH/s, but there are a LOT of those cards out there. People don't traditionally look at nVidia's gear for mining, but if there price was right there would be a lot of it out there.

Still, I think the inertia of the system would take a couple/few months to really respond to a step increase in price.

 

[Elfear]

Yes. You must assign them to separate clients since they rely on different platforms.

Thanks!

 

[Shmee]

To any moderator following along here...

This thread has 30,000 views, is 19 pages long and doesn't look like it's going to die anytime soon.

I'm not sure if the OP (thanks for posting here metalliax!) knew this would happen but do you think its time we get our own section or subsection on cryptocurrency or cryptomining?

Sure a lot of what we're talking about is video card related (as that is the current vessel for generating coins) but the zeitgeist of the thread is more about optimizing mining, mining pools, mining tweaks, tips and tricks, Ethereum speculation etc.

Just a suggestion.

A good suggestion, I have thought about it before. I will bring this up, see what ppl think.

 

[poofyhairguy]

Wow, the deals on AMD cards have dried up recently. I feel like an idiot selling my 970 when I did because now I don't have any card worth mining with.

Two questions for the group:

1. (theoretical) Is there a chance- especially if Nvidia won't have new cards until September- that the success of mining could push back the release for Polaris 10/11?

2. (practical) There are still some deals on 2GB/4GB 380 non-Xs. If I bought one of those today and mined the crap out of it until June is there a chance I could earn back its value prior to Polaris's launch? My power is medium expensive, I pay 11.04¢/kWh.

 

[MrTeal]

Wow, the deals on AMD cards have dried up recently. I feel like an idiot selling my 970 when I did because now I don't have any card worth mining with.

Two questions for the group:

1. (theoretical) Is there a chance- especially if Nvidia won't have new cards until September- that the success of mining could push back the release for Polaris 10/11?

2. (practical) There are still some deals on 2GB/4GB 380 non-Xs. If I bought one of those today and mined the crap out of it until June is there a chance I could earn back its value prior to Polaris's launch? My power is medium expensive, I pay 11.04¢/kWh.

1. I doubt it. Polaris has a lot more riding on it for AMD than some mining card sales. If AMD was really interested in mining, I think you might see more specialized mining solutions coming out.

2. I would say a very good chance, but that's just me.

 

[DrBombcrater]

2. (practical) There are still some deals on 2GB/4GB 380 non-Xs. If I bought one of those today and mined the crap out of it until June is there a chance I could earn back its value prior to Polaris's launch? My power is medium expensive, I pay 11.04¢/kWh.

At today's price ($10 per ETH) and difficulty a 380 will earn around $90 a month, after power costs. But the difficulty level has doubled in the last month and seems set to keep rising rapidly, so I'm not sure you'd ever recoup the cost of the card.

I bought my last mining card about three weeks ago, and won't be getting any more. I can't see a scenario, short of a massive jump in the price of eth (not impossible, of course), where it would pay off to buy new cards now.

 

[poofyhairguy]

Thank you. I will just keep my powder dry till Polaris then.