CryptoCurrency Mining with GPUs *spawn*

[entity279]

AFAIK, Vega has no compute switch in either Adrenalin or older drivers (either they try to auto-detect the workload or they think Vega doesn't need it at all). I also did 2-3 reinstalls of Adrenalin and 11.3, 11.4 before. RX 580 have the switch in Adrenalin, hence why with the newer Cast miner (0.8 ? ) you have flag to force the blockchain optimizations and it works for rx 580. (So for rx 580, Cast performs similarly in Adrenalin and Blockchain)

The effect of no switch is that mining on Vega works nicely on Adrenalin for ETH and poorly for XMR. I think we should try to submit a bug report, maybe they'll care.
I've given up for now and I am using the blockchain drivers. Gaming on them still works somewhat (X Fire is broken with gzillions of artifacts on GTA V, but encountered no other problems so far)

As for settings, I'm definetly running my 2 Vega Liquids with 1100 HBM & power limit of -40 %. For now I, did not lowered voltages so power consumption is pretty big.

_______
Funny, with my FX 8300 I get 220 - 280 h/s. I've just bought the 7980XE , so it must be that I will be terribly dissapointed with its mining perf. But you should be using as many threads as there are physical cores, as a rule of thumb . 18 seems overkill

 

[Deleted member 13524]

Thanks for the CPU tip! By using 10 threads on "even" cores (meaning 1 thread per core) I'm now close to 500 h/s on the CPU alone. I also disabled the power saving setting because it wasn't saving any power, it was just using less performance.
So at ~350W I'm getting 1700 Hash/s. Still a far cry from what I'd get with the blockchain driver and more clocks tuning (close to 2500 Hash/s I guess?), but I guess it's more profitable now?

Regardless, maybe it's just more profitable to do Monero on the CPU and Ethereum in the GPU?


Other thing is I'm mining for a pool that has a 0.3 XMR threshold payout, and I still have nothing in my wallet (after some ~10 hours at 1500H/s).
How can I calculate how many hours it'll take before I get some Monero from them?

 

[entity279]

Thanks for the CPU tip! By using 10 threads on "even" cores (meaning 1 thread per core) I'm now close to 500 h/s on the CPU alone. I also disabled the power saving setting because it wasn't saving any power, it was just using less performance.

Nice ! Well then hopefully the 7980XE will get at least 700 h/s. Threadripper achieves 1000 + , but that's life i guess.

So at ~350W I'm getting 1700 Hash/s. Still a far cry from what I'd get with the blockchain driver and more clocks tuning (close to 2500 Hash/s I guess?), but I guess it's more profitable now?

Regardless, maybe it's just more profitable to do Monero on the CPU and Ethereum in the GPU?

Well you can use whattomine.com and manually dial your rates for CryptoNight and EthHash to find out.
Also wattomine numers are really only useful as an estimation, if you would sell your crypto regularly and often.
If instead you'd like to keep it for a while, you should predict the groths of XMR & ETH instead and decide based on that

Other thing I'm mining for a pool that has a 0.3 XMR threshold payout, and I still have nothing in my wallet (after some ~10 hours at 1500H/s).
How can I calculate how many hours it'll take before I get some Monero from them?

Still whattomine is your friend. The pool choice doesn't matter too much (you just have to deduct its fee) ; your avereage income will be the same regardless.
Here's the per-coin view : http://whattomine.com/coins/101-xmr-cryptonight . Just input your hashrate and check the "est. reward field"

 

[CarstenS]

I've just bought the 7980XE , so it must be that I will be terribly dissapointed with its mining perf. But you should be using as many threads as there are physical cores, as a rule of thumb . 18 seems overkill

Nice ! Well then hopefully the 7980XE will get at least 700 h/s. Threadripper achieves 1000 + , but that's life i guess.

Yes, for XMR mining, Ryzen or, for that matter, Threadripper are much more useful. TR-1950X gets to around 1400-ish without being pushed to the max (at ~3,8 GHz to be exact). With 7980XE, I think you can get to low four-digit as well without overclocking.

Thanks for the CPU tip! By using 10 threads on "even" cores (meaning 1 thread per core) I'm now close to 500 h/s on the CPU alone. I also disabled the power saving setting because it wasn't saving any power, it was just using less performance.
So at ~350W I'm getting 1700 Hash/s. Still a far cry from what I'd get with the blockchain driver and more clocks tuning (close to 2500 Hash/s I guess?), but I guess it's more profitable now?

Regardless, maybe it's just more profitable to do Monero on the CPU and Ethereum in the GPU?


Other thing is I'm mining for a pool that has a 0.3 XMR threshold payout, and I still have nothing in my wallet (after some ~10 hours at 1500H/s).
How can I calculate how many hours it'll take before I get some Monero from them?

What CPU are you using with 10 threads on AMD? Don't oversubscribe, Monero loves to fit into L2-cache.

Regarding your pool: Maybe they're having problems atm? Normally, you should see a progress-bar with 9 digits behind the decimal point, so anything you mine should starting to show up after a couple of minutes, half an hour max.

 

[Deleted member 13524]

What CPU are you using with 10 threads on AMD? Don't oversubscribe, Monero loves to fit into L2-cache.

Not AMD. It's a Xeon E5 2670 v2 in a X79 motherboard.
These CPUs are mighty cheap on ebay, and I got one because I did/do lots of FES simulations on Solidworks. You can buy them for as little as $110 nowadays.

Regarding your pool: Maybe they're having problems atm? Normally, you should see a progress-bar with 9 digits behind the decimal point, so anything you mine should starting to show up after a couple of minutes, half an hour max.

I guess I'm going to change my pool then. Not willing to mine for free for anyone, problems or not.

 

[BRiT]

Not AMD. It's a Xeon E5 2670 v2 in a X79 motherboard.
These CPUs are mighty cheap on ebay, and I got one because I did/do lots of FES simulations on Solidworks. You can buy them for as little as $110 nowadays.

I picked up 2 of them for $90 a piece when Facebook upgraded their servers and they were first dumped on EBay as server pulls. I also got 128 GB of ECC server memory for nearly as cheap to go along with a dual-socket server grade motherboard. It was the cheapest server I've ever built, and all at an impulse buy because of how amazing the performance per dollar is. It's still working great on my server.

 

[Mize]

So you're mining XMR with CPUs?
I plugged that hash rate into what to mine and it thinks you should be mining SUMO or electrsomething.

 

[CarstenS]

Not AMD. It's a Xeon E5 2670 v2 in a X79 motherboard.
These CPUs are mighty cheap on ebay, and I got one because I did/do lots of FES simulations on Solidworks. You can buy them for as little as $110 nowadays.

I guess I'm going to change my pool then. Not willing to mine for free for anyone, problems or not.

Ah, I see. 500 on an older-ish 10-core is not bad in this case. Today I tried tinkering for 10 minutes on a i7-6800K, but results were not that good (sub 400, despite OC) so i quit.
I'm using nanopool.org for my experiments, no problems 'til now. But you can always get nicehash'ed I guess.

So you're mining XMR with CPUs?
I plugged that hash rate into what to mine and it thinks you should be mining SUMO or electrsomething.

Technically, yes. But (for me) it's also a question of administrability (do i have to create accounts all over the place with my banking details in them for each and every coin? Or do I limit myself to what's traded already at the exchange I'm already using) and of course a question of trust into that specific coin.

 

[A1xLLcqAgt0qc2RyMz0y]

Thanks for the CPU tip! By using 10 threads on "even" cores (meaning 1 thread per core) I'm now close to 500 h/s on the CPU alone. I also disabled the power saving setting because it wasn't saving any power, it was just using less performance.

You can improve your CPU performance even more by fully utilizing the 25MB L3 cache.

With your current setup you are only using 20MB of the 25MB L3 cache.

10 cores * 2MB cache = 20MB used.

In the CPU setup change two of the cores to "low_power_mode" : true.

Those two cores will now use two threads on the core and double the cache usage from 2MB to 4MB. The performance on those cores won't double but be about 75% higher than before.

2 cores ("low_power_mode" : true) * 4MB + 8 cores ("low_power_mode" : false) * 2MB = 24MB cache used.

Another option to save some power but still have strong H/s is to only use 6 of the ten cores and set all six of those to "low_power_mode" : true.

6 cores ("low_power_mode" : true) * 4MB cache = 24MB cache used.

Be sure not to overflow the cache as you would see drastic H/s reductions.

Remember ("low_power_mode" : true) uses 4MB cache per core and ("low_power_mode" : false) uses 2MB cache per core.

 

[Deleted member 13524]

You're right, thanks!

I simply changed 2 cores to "low_power_mode" and those two went from ~40-45H/s to 65-70H/s.
Though power consumption also went up 12-16W.

With 2 cores at 4MB cache and 8 cores at 2MB cache I was doing about 490 H/s at ~410W at the wall (this also has a Vega 64 pushing ~1830 H/s).
With 6 cores at 4MB cache I'm doing around 435 H/s at 380W at the wall.
I don't think these 55 H/s are worth 30W (I could just put an older smartphone doing that for 5W or less..) so I'm sticking with the 6 cores. I also get a rather snappier performance when I'm using my desktop with the 6 cores so that's another plus for me.



Speaking of smartphones, does someone know of an app one could use to push a couple of H/s on older smartphones and tablets? All the ones I've seen are hopelessly connected to private pools with their own wallets with ridiculous payouts that won't ever give you anything unless you have 300 smartphones mining at the same time. (i.e. they're scams..).

 

[ahu]

You're right, thanks!

I simply changed 2 cores to "low_power_mode" and those two went from ~40-45H/s to 65-70H/s.
Though power consumption also went up 12-16W.

With 2 cores at 4MB cache and 8 cores at 2MB cache I was doing about 490 H/s at ~410W at the wall (this also has a Vega 64 pushing ~1830 H/s).
With 6 cores at 4MB cache I'm doing around 435 H/s at 380W at the wall.
I don't think these 55 H/s are worth 30W (I could just put an older smartphone doing that for 5W or less..) so I'm sticking with the 6 cores. I also get a rather snappier performance when I'm using my desktop with the 6 cores so that's another plus for me.

That's interesting. I just came back to CPU mining after a hiatus of 2 years and have the exact same Xeon as you, two of them actually.

Do you have hyperthreading disabled? With my system I was doing close to 900 H/s with 20 cores and both hyperthreading and low power mode disabled. Looking forward to see how this would compare to my newly ordered Titan V

 

[Grall]

Speaking of smartphones, does someone know of an app one could use to push a couple of H/s on older smartphones and tablets?

I suggest you don't, as running the SoC full tilt over a prolonged period of time could heat damage the battery and cause swelling (and potentially a lot worse...)

It's already been seen with coinmining malware hijacking smartphones.

 

[Deleted member 13524]

I suggest you don't, as running the SoC full tilt over a prolonged period of time could heat damage the battery and cause swelling (and potentially a lot worse...)

What I have around at home is:
- HTC Butterfly 3 with a Snapdragon 810 and damaged capacitive panel (no use as a phone) and battery running out quickly
- Redmi Note 3 Pro with a Snapdragon 650 and shattered screen and broken chassis (though still fully functional).
- Redmi 4X with Snapdragon 435, though this one I'm still using as main phone while my Axon 7 is RMA'd.
- Redmi 1S with Snapdragon 400 whose battery doesn't charge anymore, only works connected to the wall.
- HTC One with Snapdragon 600 with damaged capacitive panel.
- Honor 6 with kirin 920 (4*Cortex A15 + 4* Cortex A7) with shattered screen.
- 2* Tegra 2 tablets (I don't think these will ever work though as they don't even have NEON instructions)

I was thinking the Butterfly 3 with the S810 beast I obviously have to seriously throttle it down to 1GHz or so. This isn't a problem because the device is already rooted and I've done it before.
The S435 isn't capable of warming up at all with the CPU alone. It only warms up if I'm pushing from the 4G modem + GPS + CPU. The S400 model can just get an underclock to ~1GHz and cal it a day. The Honor 6 might need some tweaks. My HTC One with the S600 only ever got warm when pushing from the GPU.

 

[Grall]

Geesh, that's a monkeyload of smartjunk you got there.

Anyway, sure, go ahead if you want to, but I'd keep the phones running well away from anything combustible if I were you... Put 'em in an oven deep pan or such, and have a fire extinguisher handy.

 

[Malo]

That’s just getting sad. Now we’re scavenging everything around the house with a cpu to mine?

 

[Deleted member 13524]

Geesh, that's a monkeyload of smartjunk you got there.

Let's just say my wife hasn't been terribly great at taking care of smartphones... Otherwise she could still be perfectly happy with that Honor 6 or the Note 3 Pro.
The Butterfly 3 was an e-bay purchase in the spur of the moment, it only cost 180€ but unfortunately only lasted a year or so before hardware problems started to appear. Before that I owned my HTC One for almost 3 years!

That’s just getting sad. Now we’re scavenging everything around the house with a cpu to mine?

Well it's either that or turning up the AC to keep the house warmer while leaving those devices to rot.
I can't sell them because they're all broken, so at least let them be of use for something.

 

[Malo]

Soon your lazy family members will be put to work solving block chains! At least let them be of use for something!

 

[A1xLLcqAgt0qc2RyMz0y]

You're right, thanks!

I simply changed 2 cores to "low_power_mode" and those two went from ~40-45H/s to 65-70H/s.
Though power consumption also went up 12-16W.

With 2 cores at 4MB cache and 8 cores at 2MB cache I was doing about 490 H/s at ~410W at the wall (this also has a Vega 64 pushing ~1830 H/s).
With 6 cores at 4MB cache I'm doing around 435 H/s at 380W at the wall.
I don't think these 55 H/s are worth 30W (I could just put an older smartphone doing that for 5W or less..) so I'm sticking with the 6 cores. I also get a rather snappier performance when I'm using my desktop with the 6 cores so that's another plus for me.

Your Hash rates still seem a little low to me.

I have a Dell T5600 with dual hex-core Xeon E5-2640 v1's. The E5-2640 has 15MB L3 cache so on each E5-2640 I use four of the six cores. Three in "low_power:true" mode (avg 86.4 H/S) and one at "low_power:false" (avg 51.4 H/S) for 14MB of L3 cache usage. I am getting 621 H/s from those eight cores. A bonus in only running on four of the six cores is that they run at 2900 MHz instead of 2800 MHz if I were to use five or six cores and as you state the system is snappier as two cores are available for general use.

http://www.cpu-world.com/CPUs/Xeon/Intel-Xeon E5-2640.html

Your E5-2670 V2 should be running at 2900 MHz

http://www.cpu-world.com/CPUs/Xeon/Intel-Xeon E5-2670 v2.html

I would expect that you should be getting around 518 H/s instead of the 435 you are seeing.

One area that may cause the lower hashes is if "Large pages on Windows" is not enabled.

I followed this guide https://getmonero.org/resources/user-guides/mine-to-pool.html to enable Large pages.

By default we will try to allocate large pages. This means you need to "Run As Administrator" on Windows You need to edit your system's group policies to enable locking large pages. Here are the steps from MSDN

1. On the Start menu, click Run. In the Open box, type gpedit.msc.
2. On the Local Group Policy Editor console, expand Computer Configuration, and then expand Windows Settings.
3. Expand Security Settings, and then expand Local Policies.
4. Select the User Rights Assignment folder.
5. The policies will be displayed in the details pane.
6. In the pane, double-click Lock pages in memory.
7. In the Local Security Setting – Lock pages in memory dialog box, click Add User or Group.
8. In the Select Users, Service Accounts, or Groups dialog box, add an account that you will run the miner on
9. Reboot for change to take effect.

 

[A1xLLcqAgt0qc2RyMz0y]

You're right, thanks!

I don't think these 55 H/s are worth 30W (I could just put an older smartphone doing that for 5W or less..) so I'm sticking with the 6 cores. I also get a rather snappier performance when I'm using my desktop with the 6 cores so that's another plus for me.

The XMR mining calculator shows it to be profitable:
https://www.cryptocompare.com/minin...Unit=H/s&PowerConsumption=30&CostPerkWh=0.075

You can plug in your own number for power cost to see if is worth it for you.

 

[Deleted member 13524]

Soon your lazy family members will be put to work solving block chains! At least let them be of use for something!

I just remembered I have a GTX 660 Ti laying around, and it supposedly does close to 300 H/s at 130W. And my desktop motherboard does have a bunch of free PCIe slots

Hmmm....

Your E5-2670 V2 should be running at 2900 MHz

Well.. it's an Engineering Sample and the base clocks are lower. My cores run at 2.7GHz max and I'm guessing the L3 cache is also running slower.

On area that may cause the lower hashes is if "Large pages on Windows" is not enabled.

I do have them enabled, but does this have anything to do with page file?
I don't have any page file configured for 2 reasons: I'm running low on my my C:\ drive and I have 64GB RAM so I never really need page files.
I tried to configure a page file into a RAM Drive I have, but it won't let me unfortunately.

The XMR mining calculator shows it to be profitable:
https://www.cryptocompare.com/mining/calculator/xmr?HashingPower=55&HashingUnit=H/s&PowerConsumption=30&CostPerkWh=0.075
You can plug in your own number for power cost to see if is worth it for you.

Yes it is. Thing is I'm using my desktop to work right now. I'm not doing heavy work, just opening word/pdf documents and occasionally compiling LaTeX, but I do need it to be snappier and using all 10 cores was making my life a bit harder/slower.

Thanks for the tips though!