Ethereum GPU mining?

Page 32 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
I see. So it has the 390, some other machine (gaming box I assume) has the 295x2 and the 7970 is wasted.

Sorry, I wasn't clear. The primary rig is i7-6700K with R9 295X2 + R9 390 in Tri-Fire. :D

That is a shame, Tahiti is a mining champion.

The 7970 is the left over card from my 7970 CF before that. Still works perfectly -- about 70*C at 1150mhz overclock. Both of those 7970 cards are from early 2012.

I get not wanting to spend more, I have three rigs running now to avoid having to upgrade a PSU to consolidate two into one. All those extra costs beyond the GPU screw up the payoff equation.

You make a good point. I might consider getting a 1000W PSU and just putting 3-4 Polaris 10 cards in it. The upfront cost is more but it's 1.5-2x the mining rate a month. I am thinking about it since it would be easy to buy a Z170 board and Skylake and resell those in 6 months given that they don't lose much value.

Good thing about Nvidia cards is their value bottoms out at a higher place on average.

Ya but that would push me even more towards expanding with more R9 390s/Polaris 10 cards than buying a $300 USD 980. I guess I'll have to do the math in Excel to see what's better over the next 6 months but it's kinda hard to estimate too since I have no idea if Polaris 10 will have hashing rate close to R9 390s or closer to Tonga's which is barely better than a 970 OC.
 

VeryCharBroiled

Senior member
Oct 6, 2008
387
25
101
I just tried the 15.12 DLLs in the same folder with the new Genoil-ethminer, and can confirm reported hashrates have gone back up. Currently on 16.4.2

Ill have to try that with the original ethminer. my r9 390 is in my daily driver/gamer rig so keeping the video driver current is a consideration. I seem to remember doing similar things during btc/cgminer times when cgminer was still on gpus.

been running the stock ethminer as its been bullet proof so far. use eth-proxy for failover and stratum. dont mind the manual DAG cleanup as its only done every week or two. "if it works.."

cant see running CDM as I have a good mining rate vs power/heat/noise at the moment. rig is in my living room so sound is pretty much number one priority. WAF ya know :)
 

poofyhairguy

Lifer
Nov 20, 2005
14,612
318
126
The 7970 is the left over card from my 7970 CF before that. Still works perfectly -- about 70*C at 1150mhz overclock. Both of those 7970 cards are from early 2012.

Wow. Too bad you can't use it, the 7970 would give a 980 a run for its money I bet. I love the 280x for mining, it is a solid 20.5 average. At that rate you can pay off a Newegg refurb one within 4 or so months. That is pretty solid, too bad Newegg is sold out for now.

With the rumors that Polaris 10 might be delayed until October I might just pull the trigger on a 280x if they come back and a new PSU.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
I would suggest using qtminer. I got it to work VERY easily on Linux.

ethminer works just fine . . . it's just that Linux is being goofy about running ethminer in terminals as different users. I was hoping to work that out so that if I had an individual card crash, it would be easier to isolate which one was causing problems.
 

zagitta

Member
Sep 11, 2012
27
0
0
ethminer works just fine . . . it's just that Linux is being goofy about running ethminer in terminals as different users. I was hoping to work that out so that if I had an individual card crash, it would be easier to isolate which one was causing problems.

Linux works perfectly fine is you just set it up properly.
Follow this guide to setup a minimal ubuntu install with drivers and XOrg configs and then just install ethminer or qtminer: https://hashcat.net/wiki/doku.php?id=linux_server_howto

I've got a 7990 running with no problems what so ever this way.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
Zagitta, did you read my earlier posts? That link does nothing to address any of my problems.
 

poofyhairguy

Lifer
Nov 20, 2005
14,612
318
126
I told myself I wasn't going to buy any GPUs before Polaris, but mining is too much fun. If anyone is looking for some GPUs that should pay off in less than a half year these refurb Tri-X 290s for $199 shipped are a great deal:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814202218&ignorebbr=1

I bought two myself. I will run those in a crossfire (always wanted to say that) in my gaming rig, then demote my 390x to pure mining use or maybe sell it while it has any value.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
That's a good buy, though I can tell you from personal experience that dealing with a lot of those reference coolers at once is hella noisy.
 

Erenhardt

Diamond Member
Dec 1, 2012
3,251
105
101
What is going to happen to ETH price after first price hike for DAO?
https://daohub.org/index.html

I'm not sure I want to invest in DAO, which seems like a quick way to increase your ETH value (+50% in 10 days).

I guess everyone that wanted to invest in DAO did that with 1.0 price rate and There will be none, or a close to 0 DAO investments in the next days. And even if someone will be buying DAO, he would do that in the exchange for better price.
The first wave of new DAO investors will arrive with first dividends from successfully kickstarted projects. Which would be at least a couple of months or even years from now.

I'm don't think I'm interested in such a long time investment.

Your opinions?
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
It's a question of whether or not the DAO will outperform the currency. Obviously those who launched Ethereum in the first place had something like the DAO in mind from the beginning.
 

Feld

Senior member
Aug 6, 2015
287
95
101
Just thought I'd mention I decided to give the claymore miner a try this weekend and it's a significant improvement for my Hawaii cards. Using ethminer, either direct or through eth-proxy, was giving me about the same results either way - reported hashrate of 28-34 MH per card, but long term effective averages more like 23-28. I'm running 290, 290x, and 390 cards, each with different clocks according to where they're stable, but now the hashrates are right in line with the reported numbers. So it's been a nice big 15%+ boost in eth production. Well worth the minor 1% dev fee. Also, the claymore miner does make the GPUs run a bit hotter than ethminer, so I've reduced my clocks a bit to compensate. I'm still averaging over 30 MHs per card.
 

MrTeal

Diamond Member
Dec 7, 2003
3,554
1,658
136
I might give the claymore miner a try on an outside system, even though I don't plan on mining the other alt-coin.

The slowdown in difficult has been much appreciated as well. I did a 3 day sliding window on the etherscan.io hashrate chart to smooth it a little bit, and from March 15 to April 15 hashrate rose 803 GH/s or 64%. In the last month though not only has the exponential growth continued, but it hasn't even been steady linear growth. Hashrate's only gone up 629GH/s or 30.5%. It's still quite profitable, and if firesales on 300 series hardware start coming more it might be worth buying a few more cards.
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
Yeah I'm starting to see 290's for cheap again. I expect 390's to drop next which depending on how well Polaris cards mine may be worth scooping a bunch of these older cards. Unlike Bitcoin there's little fear of ASICS arriving so the hashrate will see steady but fairly linear increase in difficulty. Only issue is PoS but no date has been set yet. My problem is now one of power/heat. I have run out of room to mine in my house :)
 

MrTeal

Diamond Member
Dec 7, 2003
3,554
1,658
136
Yeah I'm starting to see 290's for cheap again. I expect 390's to drop next which depending on how well Polaris cards mine may be worth scooping a bunch of these older cards. Unlike Bitcoin there's little fear of ASICS arriving so the hashrate will see steady but fairly linear increase in difficulty. Only issue is PoS but no date has been set yet. My problem is now one of power/heat. I have run out of room to mine in my house :)

I've thought about it, but I'm having a hard time making the numbers work out even with relatively cheap (CDN$250) 290s. I was thinking of setting up some tri-GPU systems, but I've burned through most of my existing hardware (other than PSUs, which I still have ~10kW sitting on the shelf). The proposed system would cost ~CDN$1000 all in, and mine ~90GH/s. The system for me would mine US~$10 day and use ~US$2.5 in power, so right now it's a net of US$7.5 for a US$800 outlay. Seems relatively safe, but I'm not sure it's really worth the time to set up and run.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
Is Claymore really that good? My Hawaii cards using ethminer do ~30 MH/s with my preferred BIOS settings at a power draw of maybe ~200W for the 290s and a bit more for the 390. One of my 290s is being stupid and only does maybe 26 MH/s @ 925 MHZ GPU (pulls ~175W), and one of my 390s has yet to be tweaked so it's sucking up way too much power right now.

Anyway that's based on the numbers from nanopool and the output from ethminer. But I might give claymore a try if I get the time to set it up.

Oh, as an aside . . . the problems I had with protocol errors and such trying to launch multiple instances of ethminer under different users was completely a user authentication problem. The easiest solution (gah) seems to be to run the terminal as root via

Code:
echo password | sudo -S su - root

and then initiate the miner. So that would require making sure each user was in the sudoers file so that they could successfully execute sudo su - root. It might be a huge security gaffe but whatever.
 
Last edited:

Despoiler

Golden Member
Nov 10, 2007
1,966
770
136
I had a user contact me with a request to see if we can make my regular Fury look more like a Nano in terms of power/perf. Below are various settings I tested. This is using 16.2.1 drivers. Also keep in mine that my Fury has one block of 4 CU unlocked compared to a stock card. 4 short of a Fury X basically. Tested with a Kill-A-Watt P3.

Idle 100 w
1050/550 +50% power limit +15mw 325w 29.4 Mh/s
1000/500 +0% power limit +0mw 296w 27.3 Mh/s
1000/500 -10% power limit +0mw 296w 27.3 Mh/s
1000/500 -20% power limit +0mw 296w 27.3 Mh/s
1000/500 +0% power limit -20mw 263w 27.3 Mh/s
1000/500 +0% power limit -30mw 259w 27.3 Mh/s
1000/500 +0% power limit -40mw 256w 27.3 Mh/s
1000/500 +0% power limit -50mw 253w 27.3 Mh/s
1000/500 +0% power limit -75mw 245w 27.3 Mh/s
900/500 -50% power limit -0mw 222w 24.6 Mh/s
900/500 +0% power limit -0mw 222w 24.6 Mh/s
800/500 +0% power limit -0mw 201w 21.3 Mh/s
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
I had a user contact me with a request to see if we can make my regular Fury look more like a Nano in terms of power/perf. Below are various settings I tested. This is using 16.2.1 drivers. Also keep in mine that my Fury has one block of 4 CU unlocked compared to a stock card. 4 short of a Fury X basically. Tested with a Kill-A-Watt P3.

Idle 100 w
1050/550 +50% power limit +15mw 325w 29.4 Mh/s
1000/500 +0% power limit +0mw 296w 27.3 Mh/s
1000/500 -10% power limit +0mw 296w 27.3 Mh/s
1000/500 -20% power limit +0mw 296w 27.3 Mh/s
1000/500 +0% power limit -20mw 263w 27.3 Mh/s
1000/500 +0% power limit -30mw 259w 27.3 Mh/s
1000/500 +0% power limit -40mw 256w 27.3 Mh/s
1000/500 +0% power limit -50mw 253w 27.3 Mh/s
1000/500 +0% power limit -75mw 245w 27.3 Mh/s
900/500 -50% power limit -0mw 222w 24.6 Mh/s
900/500 +0% power limit -0mw 222w 24.6 Mh/s
800/500 +0% power limit -0mw 201w 21.3 Mh/s

My Fury X's are set to 1100/333 -20 power limit, -50mw, 33-35Mh/s

Two of them in a non ideal system (5930K, X99 platform, 32GB RAM, Samsung 950 SSD) with a 1KW platinum power supply is only pulling 525W from the wall.

This same system I added two Asus 390's at 1111/1275 -100mv so I have a total of 4 cards running off a single EVGA 1KW power supply is only pulling 920-940W from the wall. This miner does 120-130Mh :)
 

Madpacket

Platinum Member
Nov 15, 2005
2,068
326
126
What is going to happen to ETH price after first price hike for DAO?
https://daohub.org/index.html

I'm not sure I want to invest in DAO, which seems like a quick way to increase your ETH value (+50% in 10 days).

I guess everyone that wanted to invest in DAO did that with 1.0 price rate and There will be none, or a close to 0 DAO investments in the next days. And even if someone will be buying DAO, he would do that in the exchange for better price.
The first wave of new DAO investors will arrive with first dividends from successfully kickstarted projects. Which would be at least a couple of months or even years from now.

I'm don't think I'm interested in such a long time investment.

Your opinions?


I may be a little biased as I did some needed volunteer work prior to the launch of The DAO and got to know some of the core team.

These guys are for real - Slock.it. I have no doubt they have the best intentions to grow Ethereum and IMHO The DAO or something similar is required to take it to the next level. The Slock.It team is small but they're all very talented and know what they're doing.

Whether or not it'll work is another question but I was pretty confident from my interactions with them and bought in shortly after launch, well before the DAO was at even 600K. When I bought in I had no idea it would become the most successfully funded kickstarter of all time so for me this is really exciting to see it hit $120 Million worth and growing.

There's a lot of naysayers complaining how The DAO will fail but nothing's really be done like this before. After the 28th it'll be up to us token holders to come to consensus on which projects will have the best chance of succeeding so having a lot of funds to work with is crucial to try many new things.

There's still the issue of governance, laws, and legal contracts that tie the blockchain to the physical world. I think these issues could end up being a bigger challenge than hiring devs to work on new DAPPS but that's the benefit of having so many investors as collectively we'll have to figure this out. There's global expertise invested in it to help it succeed. This isn't your typical VC.

A good blog was written about the challenges that lie ahead which I recommend anyone who is interested in Slock.it's - The DAO give a read:

https://medium.com/@colony/the-dao-how-to-not-fuck-it-up-5d57f354d0e7#.ejadxg3xz

Anyways I never give investment advice so I won't recommend anyone buy in until they've done their homework but even if The DAO fails it will be a historic event. This is essentially a giant social experiment where no one really knows what'll happen or how it'll succeed but there's no shortage of ideas or opportunity to work with.

That's part of the fun though, and being able to vote for ideas you think will succeed or even coming up with your own proposals to seek funding is exciting.

I think this is why already it's successful from a funding standpoint. People are interested in trying something new, something fresh as there's too many middle men on the planet, too many hands taking pieces along the way, too much corruption in the existing structures, poor security, fraud abuse etc. This is where a Ethereum and contracts make sense and the DAO framework was created to kickstart it all.

Fun times ahead!
 
Last edited:

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
What I don't like about the DAO is that they clearly mention the code could have bugs and if you loose your money it was your risk and your fault for taking the risk. So the code might have a bug and you loose your investment or they actually planted a bug in it and will cash out themselves.

Anyway, it doesn't exactly scream high confidence from their side.
 

poofyhairguy

Lifer
Nov 20, 2005
14,612
318
126
What I don't like about the DAO is that they clearly mention the code could have bugs and if you loose your money it was your risk and your fault for taking the risk. So the code might have a bug and you loose your investment or they actually planted a bug in it and will cash out themselves.

Anyway, it doesn't exactly scream high confidence from their side.

You can look at the contract code, so there is no secrecy. I am not a programmer, but I assume someone who puts in $millions did due diligence. I bought some tokens just to be a part of history.
 
Last edited:

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
You can look at the contract code, so there is no secrecy.

Yeah sure, everyone can also look at OpenSSL source code and yet not even Google manages to find all bugs or takes a very long time.

That argument doesn't count, as I don't have the time or expertise to do it. You need to trust them and their disclaimer makes that very hard.

Would you give your money to a bank that says "And maybe our code has security issues and if we lose your money it's totally your fault".