Ethereum GPU mining?

Page 30 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

JBT

Lifer
Nov 28, 2001
12,095
1
76
Dang late to the party again, may have to spin up some of my cards for this.
 

poofyhairguy

Lifer
Nov 20, 2005
14,612
317
126
Ok, I answered my own question. There isn't much of a difference between ether-proxy and qtminer, except qtminer will try to submit rejected shares while ether-proxy won't (as a speed choice). I can't tell a difference between them in mining as far as hash-rate

Mining software that DOES make a difference to me is the claymore miner:

https://bitcointalk.org/index.php?topic=1433925.0

I am getting a reported hashrate 6% higher (we will see how it actually plays out over time) which completely covers the 2% software fee. PLUS I am now mining a little decred coin on the side on the same 390x GPU, which hopefully will cut into the power costs (I estimate 50%!). So 4% extra plus some other coin for free power? Hell yeah! Here is a full analysis of mining profitability, I am doing decred at -40 and it is not effecting my reported Eth hashrate really (like 0.5%):

https://forum.ethereum.org/discussion/6652/claymores-dual-miner-power-profit-test

One other thing I found out is this Claymore miner LOVES overclocking. Loves it. So if you have a really highly clocked GPU I would at least try it.

I also got my other machine up and running- 280x at 1050 on Ubuntu 14.14. Uses 270 watts which is nice, but my hashrate seems low right now- around 17. I was using qtminer at first to test, I will probably switch that one to the Claymore miner soon to try and get a better hash rate.
 
Last edited:

thilanliyan

Lifer
Jun 21, 2005
11,450
1,357
126
https://bitcointalk.org/index.php?topic=1433925.0

I am getting a reported hashrate 6% higher (we will see how it actually plays out over time) which completely covers the 2% software fee. PLUS I am now mining a little decred coin on the side on the same 390x GPU, which hopefully will cut into the power costs (I estimate 50%!). So 4% extra plus some other coin for free power? Hell yeah! Here is a full analysis of mining profitability, I am doing decred at -40 and it is not effecting my reported Eth hashrate really (like 0.5%):
Which driver version are you using? I don't wanna go back to 15.12 lol.
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
How does this actually work? And what is the benefit?
Do you actually get something like a dividend every once in a while? Or is it just like buying a stock that can appreciate?
I recommend everyone interested read the following. It's dated but gives the underlying principles over a lot of the terminology that's now become commonplace.

https://blog.ethereum.org/2014/05/06/daos-dacs-das-and-more-an-incomplete-terminology-guide/

Once you're done reading that watch this presentation by Christoph Jentzsch (core developer of "The DAO") who works for Slock.it.

https://www.youtube.com/watch?v=49wHQoJxYPo

It's pretty simple conceptually but there's a lot of little details still being worked out.

"The DAO" is the first Distributed Anonymous Organization "framework" created by members of Slock.it. This framework was released to the community free and as open source so anyone can fork it or extend it. However "Slock.it" the company created a fundraiser or crowdfunding which is owned by the people who invest in it, not Slock.it the company. You send the DAO "Ether" and you get back "Tokens". Think of it like shares in a company but that company is controlled by no single entity, it's decentral and controlled by the Token holders. The contract details are stored in the Ethereum Blockchain.

As for the money making part. The idea is for token holders of The DAO to choose the best decentral applications (DAPPS) and if those DAPPS are profitable dividends are paid to the shareholders. The amount paid out is dependent on the contract details but 1-5% has been proposed as a starting point.

More on the DAO Token: For one unit of Ether you get 100 tokens (at least until the first round of investment funding ends in 9 days). These tokens will likely appreciate in value provided the DAPPS chosen by the token holders are profitable. Then you have the amount of ether defined in the contracts which again is chosen by the token holders. It's consensus based voting. It all depends on the contract details but multiple DAPPS will be proposed, many will likely be accepted so there's a good chance one or two will be profitable even if 10 fail miserably.

The nice thing is the more crowdfunding money that is sent to The DAO, the larger the number DAPPS that can be accepted, or larger more ambitious DAPPS can be proposed and chosen. Currently The DAO fund is sitting around 15 Million USD with plenty of time left and is already in the top 10 for the most successful crowdfunding launches after five days! This is pretty monumental to be honest.

You may wondering why would Slock.It release The DAO framework for free and not own any part of The DAO? Well Slock.It is made up of some very talented individuals and they have their own DAPP proposal which they hope the community votes for (but don't have to).

Here's Slock.It's DAPP proposal:

https://slock.it/faq.md#a2

If you have any in depth questions there's a #Slack setup (slock.it) dedicated to discussions on The DAO and Ethereum. I encourage anyone interested join and ask away in the #general channel. Many people there to help out.
 
Last edited:

poofyhairguy

Lifer
Nov 20, 2005
14,612
317
126
How does this actually work? And what is the benefit?
Do you actually get something like a dividend every once in a while? Or is it just like buying a stock that can appreciate?
Madpacket nailed it. I see the shares as something like a lottery ticket with a payout in a decade or so if it hits.
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
Ok, I answered my own question. There isn't much of a difference between ether-proxy and qtminer, except qtminer will try to submit rejected shares while ether-proxy won't (as a speed choice). I can't tell a difference between them in mining as far as hash-rate

Mining software that DOES make a difference to me is the claymore miner:

https://bitcointalk.org/index.php?topic=1433925.0

I am getting a reported hashrate 6% higher (we will see how it actually plays out over time) which completely covers the 2% software fee. PLUS I am now mining a little decred coin on the side on the same 390x GPU, which hopefully will cut into the power costs (I estimate 50%!). So 4% extra plus some other coin for free power? Hell yeah! Here is a full analysis of mining profitability, I am doing decred at -40 and it is not effecting my reported Eth hashrate really (like 0.5%):

https://forum.ethereum.org/discussion/6652/claymores-dual-miner-power-profit-test

One other thing I found out is this Claymore miner LOVES overclocking. Loves it. So if you have a really highly clocked GPU I would at least try it.

I also got my other machine up and running- 280x at 1050 on Ubuntu 14.14. Uses 270 watts which is nice, but my hashrate seems low right now- around 17. I was using qtminer at first to test, I will probably switch that one to the Claymore miner soon to try and get a better hash rate.
Thanks for the links. Decred appears to be worth around $1.50 a coin so if you can flip that it may be worth the added power costs. I did this dual mining a while back with Bitcoin / Namecoin and made a few hundred off the NMC.

I don't buy the 5-6% performance improvements (looks more like 1-2% at best, reported hashrates are a bit of a bad measurement, effective current hashrates over a set period of time is all that really matters) but hope to be proven wrong. Can you report back in a few days and see if you actually notice a difference? also do you have any power measurement differences when mining both coins vs just ethereum?

For me the amount of downtime and tweaking to switch over 8 miners would probably lose me more money the 1-2% performance improvements (unless it really is closer to 5%). I think the real value here is in mining the Decred coin at the same time but if it increases power costs by 15-20% maybe not.

Thanks for sharing though, this could be interesting :)
 

poofyhairguy

Lifer
Nov 20, 2005
14,612
317
126
Can you report back in a few days and see if you actually notice a difference?
Can do, I will keep an eye on it. Apparently some other people are working on optimized miners, and there is as much as 30% performance sitting on the table for whoever can change the waveform. This might just be the beginning of new mining software, which will make the difficulty go up very quickly.

also do you have any power measurement differences when mining both coins vs just ethereum?
I will move the Kill-a-Watt over as soon as I finish configuring rig 2.

For me the amount of downtime and tweaking to switch over 8 miners would probably lose me more money the 1-2% performance improvements (unless it really is closer to 5%). I think the real value here is in mining the Decred coin at the same time but if it increases power costs by 15-20% maybe not.
Yeah, the value is the decred but its difficulty is rising at an insane level. Unles decred price shoots up I can see it being a viable situation for only a few months or so.
 

thilanliyan

Lifer
Jun 21, 2005
11,450
1,357
126
Okay tried out that Claymore miner linked to.
It says I'm getting 62MHs ether and 1500MHs on DCR, with 2x290 at 1100/1250. This is at DCR -50.

Power usage is up to about 680w from 550w.
According to this calculator I am making enough DCR to offset a lot of the power costs. However, my VRM temps jumped almost 20C so I don't think I'm going to keep running it.
http://decred-calc.cryptohub.info/

EDIT:
I underclocked and undervolted to 947 core and -50mV and now I'm mining at 53MHs ether and 1300MHs DCR (-50), pulling 600w and the VRM temp only went up about 6C.

EDIT2:
I underclocked and undervolted to 947 core and -100mV and now I'm mining at 53MHs ether and 1300MHs DCR (-50), pulling 560w and the VRM temp only went up about 3C.
 
Last edited:

RussianSensation

Elite Member
Sep 5, 2003
19,458
764
126
I also got my other machine up and running- 280x at 1050 on Ubuntu 14.14. Uses 270 watts which is nice, but my hashrate seems low right now- around 17.
I have a spare Sapphire Dual-X 7970 that I tried 2 days ago. @ 1125mhz on stock voltage of 1.175V, the best I got out of it was 18.5-20MH/sec, even though it benchmarked 22-24MH/sec (ethminer -G -M). However, while actually mining, it was much closer to 18.5-20MH/sec, substantially less than my Hawaii cards which are doing 30-31.5MH/sec with the same command lines. I actually thought Tahiti is much better than that but it isn't. On stock voltage, with an i7 6700K @ 800mhz, total power usage was 226W with a 1000W SeaSonic Platinum.

I am curious to know how well GP104 will mine against Polaris 10 and Hawaii. I have a feeling a $400-600 GP104 will not even be as good as Polaris 10/Hawaii but we'll see. I hope someone posts mining benches for these cards as I wanted to put together another 1-2 mining rigs.

Current supply checks on Newegg show that AMD continues to send new R9 390 cards.

MSI Radeon R9 390 DirectX 12 R9 390 GAMING = 300 new units available
PowerColor PCS+ Radeon R9 390 = 300 new units available

Looks like someone at AMD did take notice as new shipments of 200-300 units are coming in as soon as the previous shipment sells out. This also means the network rate will continue rising as every single R9 390 shipped sells out. With so many new R9 390 cards arriving, I am not even sure Polaris 10 will launch in retail around Computex May 31-June 4th. Just 600 PowerColor PCS+ and MSI Gaming 390s have to sell in a just a month if Polaris 10 is going retail June 4th unless AMD is going to be major discounts on remaining inventory.

Some R9 390X cards are selling for almost the same price as the R9 390s.
 
Last edited:

Shmee

Memory and Storage, Graphics Cards
Super Moderator
Sep 13, 2008
5,864
1,264
136
Anyone try the latest 16.5.1 drivers? Any difference from the others post 16.2?
 

DrMrLordX

Lifer
Apr 27, 2000
19,155
7,915
136
Okay tried out that Claymore miner linked to.
It says I'm getting 62MHs ether and 1500MHs on DCR, with 2x290 at 1100/1250. This is at DCR -50.
What driver version are you using? Also, is that a Windows or Linux client?

62 Mh/s is about what I would expect from 290s @ 1100 MHz, I get around 60 MH/s with two 290s @ 1075 MHz (or a 390 + 290 with same clocks).
 

poofyhairguy

Lifer
Nov 20, 2005
14,612
317
126
I have a spare Sapphire Dual-X 7970 that I tried 2 days ago. @ 1125mhz on stock voltage of 1.175V, the best I got out of it was 18.5-20MH/sec, even though it benchmarked 22-24MH/sec (ethminer -G -M). However, while actually mining, it was much closer to 18.5-20MH/sec, substantially less than my Hawaii cards which are doing 30-31.5MH/sec with the same command lines. I actually thought Tahiti is much better than that but it isn't. On stock voltage, with an i7 6700K @ 800mhz, total power usage was 226W with a 1000W SeaSonic Platinum.
It is kinda crazy how much the effective hash rate varies. Last night the 280x did 17-18, when I came home a few hours ago it was 26.7 (and my 390x was at 34.4)! Now the rate is exactly what I would expect- 30 for 390x and 20 for 280x. I will wait a few days before I move the 280x to the Claymore because from what I understand Tahiti don't do as well mining both coin types as Hawaii so it's all got to be for Eth benefits.

I am curious to know how well GP104 will mine against Polaris 10 and Hawaii.
I am not going to wait. First Polaris on Newegg at MSRP I am buying it. I think the first round will sell out just because of mining, especially if there is any fast confirmation of decent hash rates.

Some R9 390X cards are selling for almost the same price as the R9 390s.
Wow that is my card and a good deal.

Power usage is up to about 680w from 550w.
According to this calculator I am making enough DCR to offset a lot of the power costs. However, my VRM temps jumped almost 20C so I don't think I'm going to keep running it.
http://decred-calc.cryptohub.info/
Yeah I might not keep it up too long either, maybe just a month for fun. If you look at the effective rate for me it was WAY lower- 500 vs 250! Decred might not make sense, I will judge the miner on average Eth rate over the next couple of days.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
19,155
7,915
136
Using 16.2 on Windows 10.
Okay . . . the Claymore guy was pretty adamant about 15.12 so it's good to hear that some other driver versions can work with it.

Yeah I might not keep it up too long either, maybe just a month for fun. If you look at the effective rate for me it was WAY lower- 500 vs 250! Decred might not make sense, I will judge the miner on average Eth rate over the next couple of days.
Please let us know. So far Claymore doesn't seem to be much of an improvement for Ethereum, and I have my doubts about Decred. At least it isn't Dogecoin . . .
 

Staple

Member
Jun 9, 2013
30
0
0
I tried the Claymore miner...definitely better speeds on ETH but it draws too much power and gets WAY too hot to dual mine decred. Not the GPU but the on-card memory and VRMs.

Once I factor in the extra power usage, 2% fee, etc. it looks like it'd be an extra $5 monthly MAYBE if I actually get the decred predicted (which didn't seem to be happening when I tried a few pools).

Also, with so many people mining decred now the difficulty is really rising and very soon it will not be profitable at all unless the price moves beyond $1.50.
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
My issues is I've got low power margins on my miners. I'm running 4x 390's on a 1050W Seasonic Gold pulling 990W from the wall, 2 Fury X and two Asus 390's on a EVGA 1KW P2 pulling 950W. I have only one milk crate miner with a 1300W G2 that I could dual mine on. The other setups I'm already within 10-15% of Max load so would have to be very careful switching. I'm pretty convinced though I have things dialed in well enough (performance per watt). Most people I see on that thread are running only two 390(x)'s on a 1KW power supply where I'm running 4 cards no problems after tweaks. I should be able to test out on my MSI milk crate miner though as I have at least 350W of room left but I would probably be better off just adding one more 390 ;)

The added power / heat may not be worth it but only way to find out is to test.

FYI I just passed 310 coins minted with my own gear. Well on my way to 500 which was my original goal. Even with difficulty curve I should hit that by mid June or early July.

Oh and I'm definatley preordering a Polaris or Vega card. I just hope some ITX form factor sized cards are available on launch so I can replace my Asus 970 Geforce. Need to sell that quick before new cards arrive. The 970 has pulled in around 12 coins since I started so early so not bad, almost $150 CAD worth before power costs. I'll unload for 300+ and then put that money towards Polaris or Vega.

Fun times.

Keeping the 390's. They're going to be worth nothing once PoS hits as the used market will flooded but eventually dry up and the price will settle around $200-$220.
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
I tried the Claymore miner...definitely better speeds on ETH but it draws too much power and gets WAY too hot to dual mine decred. Not the GPU but the on-card memory and VRMs.

Once I factor in the extra power usage, 2% fee, etc. it looks like it'd be an extra $5 monthly MAYBE if I actually get the decred predicted (which didn't seem to be happening when I tried a few pools).

Also, with so many people mining decred now the difficulty is really rising and very soon it will not be profitable at all unless the price moves beyond $1.50.
Yeah this is my fear too. Also what are you paying for electricity?

Remember every minute of downtime is lost revenue and this scales proportionally the more Mh you are mining at. I haven't done the math but if it takes me roughly two-three days to tweak 9 miners is that worth switching over given the amount of downtime necessary to dial in the proper settings using Claymore? If on average I generate 2-3% more per miner how long would that take to even break even with the downtime. Also factor in the difficulty curve during dowtime and added power costs etc.
 
Last edited:

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
I have a spare Sapphire Dual-X 7970 that I tried 2 days ago. @ 1125mhz on stock voltage of 1.175V, the best I got out of it was 18.5-20MH/sec, even though it benchmarked 22-24MH/sec (ethminer -G -M). However, while actually mining, it was much closer to 18.5-20MH/sec, substantially less than my Hawaii cards which are doing 30-31.5MH/sec with the same command lines. I actually thought Tahiti is much better than that but it isn't. On stock voltage, with an i7 6700K @ 800mhz, total power usage was 226W with a 1000W SeaSonic Platinum.

I am curious to know how well GP104 will mine against Polaris 10 and Hawaii. I have a feeling a $400-600 GP104 will not even be as good as Polaris 10/Hawaii but we'll see. I hope someone posts mining benches for these cards as I wanted to put together another 1-2 mining rigs.

Current supply checks on Newegg show that AMD continues to send new R9 390 cards.

MSI Radeon R9 390 DirectX 12 R9 390 GAMING = 300 new units available
PowerColor PCS+ Radeon R9 390 = 300 new units available

Looks like someone at AMD did take notice as new shipments of 200-300 units are coming in as soon as the previous shipment sells out. This also means the network rate will continue rising as every single R9 390 shipped sells out. With so many new R9 390 cards arriving, I am not even sure Polaris 10 will launch in retail around Computex May 31-June 4th. Just 600 PowerColor PCS+ and MSI Gaming 390s have to sell in a just a month if Polaris 10 is going retail June 4th unless AMD is going to be major discounts on remaining inventory.

Some R9 390X cards are selling for almost the same price as the R9 390s.

600 390's will sell pretty quickly with the earnings of mining ether at the moment. The DAO is doing a lot to keep prices per Ether up and I suspect the price will peak near day 28 of the offer. Already over 2% of all ether ever generated or sold from the premine has been committed to the DAO within 5-6 days. I bet NewEgg is making good margins on these cards as given how close to next gen launch we are. The DAO is on track to being in the top 3 of all time funded crowd funding.


https://en.m.wikipedia.org/wiki/List_of_highest_funded_crowdfunding_projects
 
Last edited:

ASK THE COMMUNITY