• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Ethereum GPU mining?

Page 23 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Madpacket said:
I know you have Elpedia but my Hynix cards perform in the same manner. Hynix vs Elpedia and higher memory speeds made a difference for Bitcoin though. I set all my memory speeds to 1275Mhz and get almost the same performance as setting up to 1600. The nice thing is that power consumption drops quite a bit allowing you to spend your electrical budget on GPU speeds.

Also IIRC 290's and likely 390's have "memory ramps". It would be nice to get a list of those ramps so we can adjust speeds accordingly. Need to look closer at Hawaii BIOS editor.
Click to expand...
Same thing here. I'm mining with 2x 290, 2x 290x, and 2x 390. Memory clocks appear to make no discernible difference in hash rate from 1200 to 1650 MHz. Core clock makes a big difference, from 29-30 MH at 1040 MHz to ~34 at 1230 MHz.
 
Memory clocks on my 390 and 290s (testing four of them now) do nothing but "even out" the reported hash rates while mining with a high global work setting. The average doesn't change much.

edit: it's a shame that more work hasn't been done to optimize for newer cards. Isn't the source for stuff like Ethminer open source? Couldn't we um . . . optimize it ourselves?
 
RussianSensation said:
Ok I literally just went to my computer and tested it.

R9 390
1001 mhz = 25.3-28.2 MH/sec (mininum - maximum range)
1165 mhz = 30.7-33.8 MH/sec
1220 mhz = 33.5-36.5 MH/sec

The flag I am using is
"--cl-local-work 256 --cl-global-work 8192"

Changing memory speed from 1500mhz to 1600mhz has no material impact on performance at any of those GPU clocks.

Are you sure your R9 390/390X card's overclock is actually being applied?
Click to expand...

Interesting, I'll test it out tomorrow, it's at a remote location and I don't like messing with the clocks too much when I don't have physical access. The clocks may not be applying correctly on that box.
 
Last edited:
DrMrLordX said:
edit: it's a shame that more work hasn't been done to optimize for newer cards. Isn't the source for stuff like Ethminer open source? Couldn't we um . . . optimize it ourselves?
Click to expand...

Yeah but do you know any OpenCL expert? Because that's exactly what is needed to optimize for specific uArch.
 
Hah! I've found that waiting for AMD to do anything on the software side can be hit-or-miss. Ask anyone who's been following HSA development . . . bleh.

But hey, the optimization should be trivial for someone of the calibre necessary to work for AMD's RTG. Or so I would think.
 
thilanliyan said:
Which driver version?
Click to expand...

I took the advice on 16.2.1.

RussianSensation said:
:thumbsup: R9 390X > 970, more VRAM, better DX12 support, and now it starts to pay for itself. Win-Win.
Click to expand...

Thanks RS. I would feel a lot smarter about it if I would had just bought a 290x in the first place lol. I learned my lesson, I will try to buy a Polaris 10 card the second it hits Newegg. If your theory is right and the 390x is a better miner than Polaris I might just throw the 390x in some rig in the closet until either Ethereum either becomes worthless or too difficult to mine with GPUs.
 
Ya I went back to 16.2 as it looks like all the later ones have worse hash rates. Hopefully DOOM beta works fine with them.
 
poofyhairguy said:
I would feel a lot smarter about it if I would had just bought a 290x in the first place lol.
Click to expand...

Reference 290s (and maybe some of the non-reference 290s) are pretty sweet if you're willing to tinker with old, funky cards. I have reference XFX and Gigabyte cards doing ~29-30 MH/s using about 30mv less than what my MSI 390 Gaming takes to do about the same.

390: 1072 GPU clock, 1500 mem clock, 1.094v VDDC reported by GPU-z
XFX/Gigabyte 290: 1075 GPU clock, 1500 mem clock, 1.063v VDDC reported by GPU-z

Those two 290s have reference coolers that aren't complete ass, though they are still a bit noisy at 70% fan on upwards, and that's in a room with some serious fan noise already. The 390 is basically silent. And caveat emptor on 290s since they may have been around the block a few times, have BIOS issues (bleh), dried out/inadequate TIM applications on the GPU, and other nonsense. I have apparently run into both on some eBay Sapphire reference 290s.

In any case, I don't see a lot of point to getting the 390x/290x over the 390/290 since you will get maybe 2-3 MH/s that way, and you'll burn more power to get it.

I learned my lesson, I will try to buy a Polaris 10 card the second it hits Newegg. If your theory is right and the 390x is a better miner than Polaris I might just throw the 390x in some rig in the closet until either Ethereum either becomes worthless or too difficult to mine with GPUs.
Click to expand...

Polaris should be interesting. I don't see how the 390x will wind up being the better mining card unless software support for it just isn't "there". It should be incredibly efficient compared to Hawaii.
 
DrMrLordX said:
In any case, I don't see a lot of point to getting the 390x/290x over the 390/290 since you will get maybe 2-3 MH/s that way, and you'll burn more power to get it.
Click to expand...

Yeah I wouldn't buy an X version for mining. I got my 390x because:

1. It was cheap ($309) which will help with eventual recouping that value via a resale

2. If Polaris disappoints (aka is below a 390x at 1080p for some ungodly reason) I am set for longer.

Polaris should be interesting. I don't see how the 390x will wind up being the better mining card unless software support for it just isn't "there". It should be incredibly efficient compared to Hawaii.
Click to expand...

Polaris will have half the memory bus bandwidth (512vs 256) and fewer shaders than Hawaii. I could see Polaris rocking Hawaii in games (especially at 1080p) but falling behind in mining. We will see.
 
Madpacket said:
I almost bought that Devil 13 a few weeks ago, hesitated for a bit and then it was gone 🙁 They had it with a $100 rebate for a reasonable price a NewEgg.ca. I figured without watercooling like the 295X2 reference cards it would get toasty. Still nice to have that my MH's on one card. Trying to find a 295X2 second hand but can't find anything under $700.00 CAD which to me isn't really worth it. For $600 I would bite for a second hand card like this.
Click to expand...

Just an update on this. The rebate got approved so I just need to do the 6 week wait, and I sold the mouse that came with it locally for $120. That makes my all in cost on the 390 II CDN$740 with shipping and 5% GST. That's a good bit better per core than the XFX 390s ended up being as they cost $438 (after $20MIR). I still got a Hitman code out of the 390, but the value of that is questionable as they're pretty hard to offload for a reasonable price.
 
Ethereum price is still in free fall it seems. I think a rebound in the long term is likely, but maybe this will cool off the sharp difficulty increases.
 
MrTeal said:
Just an update on this. The rebate got approved so I just need to do the 6 week wait, and I sold the mouse that came with it locally for $120. That makes my all in cost on the 390 II CDN$740 with shipping and 5% GST. That's a good bit better per core than the XFX 390s ended up being as they cost $438 (after $20MIR). I still got a Hitman code out of the 390, but the value of that is questionable as they're pretty hard to offload for a reasonable price.
Click to expand...

Wow, that's amazing. The mouse was worth $120? I didn't notice that when looking at the card. Awesome.
 
Madpacket said:
Yeah it's kinda rough watching it fall like this. I think everyone should read the following thread about software development for Ethereum.


https://www.reddit.com/r/ethtrader/comments/4egj2g/some_thoughts_for_the_impatient_from_a_software/
Click to expand...

That is kinda dense. The laymens point I got out of that is "it will be a while before Eth is of practical use by regular folks."

I think that is overrated Mind you regular people using it is needed in the long run. But they aren't the first movers. Here is a great read on the subject:

https://www.equities.com/news/how-bitcoin-made-it-past-the-black-market
 
Last edited:
Played with the clocking today and it seems you guys are right. OCing through the AMD drivers produces significantly different results than I was getting from Afterburner. Whoops.

poofyhairguy said:
That is kinda dense. The laymens point I got out of that is "it will be a while before Eth is of practical use by regular folks."

I think that is overrated though. I hate to say it, but bitcoin isn't a thing because of legitimate people using it to buy coffee. It is a think because the criminal element of the world is quite a massive market when its all put together, and they needed a currency to facilitate their activities not tied to one government entity. I think when the practicality of bitcoin drops for these people they will move to Eth not because of the profit potential, but just to get business done. They aren't doing it for currency speculation, they are doing it for the benefits to money laundering, drug purchasing, etc.

Mind you regular people using it is needed in the long run. But they aren't the first movers. Here is a great read on the subject:

https://www.equities.com/news/how-bitcoin-made-it-past-the-black-market
Click to expand...

There is a dark market DAO in development which, I hate to say, will probably increase ETH value faster than any of the other DAPPS.
 
Last edited:
poofyhairguy said:
That is kinda dense. The laymens point I got out of that is "it will be a while before Eth is of practical use by regular folks."

I think that is overrated Mind you regular people using it is needed in the long run. But they aren't the first movers. Here is a great read on the subject:

https://www.equities.com/news/how-bitcoin-made-it-past-the-black-market
Click to expand...

I'm not sure what an article on bitcoin from 2014 has to do with Ethereum development?

If the message you're trying to get across is a currency needs drugs or something similar to make it valuable sure I get that. But you realize Ethereum is not Bitcoin right? It's a decentral Turing complete computer that's vastly more advanced than Bitcoin will ever be. This of course comes with tremendous complexity as the reddit dev points out. A new way of thinking, real programmers, not hacks will be able to wing it and make contracts. These applications will take a long time so I hope everyone here is going long and holding.

The utility is now left to the imagination of competent developers. This freedom and potential is what makes it so valuable and different than Bitcoin. Bitcoin is great for currency exchange, multi-sig and a few other commerce related activities and can easily co-exist with Ethereum due to how different they are.
 
EightySix Four said:
Played with the clocking today and it seems you guys are right. OCing through the AMD drivers produces significantly different results than I was getting from Afterburner. Whoops.



There is a dark market DAO in development which, I hate to say, will probably increase ETH value faster than any of the other DAPPS.
Click to expand...

With the capability of Ethereum over Bitcoin you may be right if they use it to do something new. I'm looking forward to completely new apps that haven't been done before.
 
Madpacket said:
If the message you're trying to get across is a currency needs drugs or something similar to make it valuable sure I get that.
Click to expand...

It needs a killer application like bitcoin gained from certain illegal activities or VHS got with porn. I don't expect it to be the same as bitcoin and I don't expect one to replace the other. The difficult nature of Ethereum development is a unique opportunity and I see a lot of advantages for whatever organization (criminal or otherwise) develops the killer app.
 
EightySix Four said:
There is a dark market DAO in development which, I hate to say, will probably increase ETH value faster than any of the other DAPPS.
Click to expand...

That makes sense. There is a lot of incentive to create such a DApp as you can make a ton of money from it.

However a killer App means nothing if it isn't easy to use. And easy-to use for me includes auto-backup and restore of Ethereum Addresses ("wallets"). You can't expect users to backup their own keystore. That just won't work as most won't do it. Even if it is their money on the line.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top