Thanks :-)

Doing it manually is an option, but I was hoping there would be a more automatic solution that doesn’t require people to edit each rig they have.

Maybe the SMOS developers can add some command line switch that does the same thing during compiling.

Looking for a ” Miner OPTIONS” configuration for CryptoNight-Heavy that makes the AMD RX580 8GB cards run with their peak performance.
 
Under Windows 10 with SRBMiner these cards run around 950 H/s per card, but in SMOS they end up around 660 H/s in both XMRig and XMR-Stak, so if someone know which setting I need to add to the configuration in SMOS to solve this speed issue that would be great.

I thought it would be a good thing to gather actual payout data from people in the community and compare pools against each other on different currencies.

My suggestion is that people willing to compare pools put the same hash rate on two pools at the same time, preferably with more than 1 miner on each pool, and gather the 24h payouts for at least 7 days.

Here is my test:

Antpool vs Bitcoin.com – BCH

5 x Antminer S9 on each

17/01-18
0.05439366 - Antpool
0.05465084 - Bitcoin.com

18/01-18
0.06105985 - Antpool
0.06112462 - Bitcoin.com   

19/01-18
0.05826221 - Antpool
0.06010202 - Bitcoin.com

20/01-18
0.05925723 - Antpool
0.06023640 - Bitcoin.com

21/01-18
0.05246061 - Antpool
0.05320492 - Bitcoin.com

22/01-18
0.05476580 - Antpool
0.05659880 - Bitcoin.com

23/01-18
0.05702378 - Antpool
0.05765114 - Bitcoin.com

24/01-18
0.05482190 - Antpool
0.05390700 - Bitcoin.com

25/01-18
0.05190309 - Antpool
0.05263578 - Bitcoin.com

26/01-18
0.04964170 - Antpool
0.05084425 - Bitcoin.com

27/01-18
0.04978585 - Antpool
0.05076544 - Bitcoin.com

Bitcoin.com wins by a tiny margin vs Antpool for BCH

Primarily an ETH issue on AMD at the time, with the older drivers, but with the beta Blockchain drivers, or possibly even the manual configuration in Radeon Settings to adjust GPU Workload options to Compute instead of Graphics on the newer updates, this issue is basically gone.

If you haven't installed the DAG fix drivers from AMD your Panda will slowly drop over time on ETH.

http://search.amd.com/en-us/Pages/results-all.aspx#k=blockchain

The B3 should run stable 229MH/s on 220V with the latest software installed.

@Claymore

I see the following error now and then, after a Share has been found and where you would normally see the "Share accepted (xx ms)!" message:

XMR: 10/02/17-17:33:04 - SHARE FOUND - (GPU 2)
cannot find block data:
{"id":4,"jsonrpc":"2.0","result":null,"error":{"code":-1,"message":"Lowdifficultyshare"}}
cannot convert blob data

so what is happening here, is the share rejected from the pool or what?

Thanks!

Just because *you* buy things at ridiculously inflated prices doesn’t mean others have to, so my latest 15 x RX580 rig cost me around $4400, but it's annoying that it has to run Linux with DAG issue to get all cards hashing.

For ETH it is already available as a switch called -minspeed and it works fine for the issue you are describing, just set it to be somewhat below what you normally get from your rig.

You need to take a chill pill.

https://www.coinwarz.com/network-hashrate-charts/ethereum-network-hashrate-chart

https://www.coinwarz.com/difficulty-charts/ethereum-difficulty-chart

Seeing as you can get a full rig with 8 x RX580 for around $2000 these days in your local computer shop, with significantly better warranty and a resell value on the video cards, that would suggest that the price on the Panda B3 needs to come down in price to what it used to be when they were launched.

Other than that, the ones I got a couple of months back have been running non-stop with no problems since I started them.

Only annoying thing for me was that they came with only Chinese language configured, meaning you have to figure out which buttons to push to install English in addition.

Panda should probably also upgrade from RX470 to RX570/80 in their next batch, since I don't think the RX470 chip is cheaper for them to buy these days compared to RX570/80.

Tell them to fix the annoying 8 x AMD GPU limit issues in Win 10 while they are at it ;-)

Try adding -dcri 8 as a switch

Thanks for all the input, but what I ended up doing instead of fiddling with the WattMan/Claymore settings, just to single out that 1 card behaving differently, was just flashing the ROM of all cards with the 480LowEnergyOK posted here:

https://bitcointalk.org/index.php?topic=1584617.0

I am using:
6 x Sapphire Radeon Nitro+ OC 8GB GDDR5 (Samsung)
#SKU: 11260-01-20G

During the flashing I found that 1 of the cards was older and had a lower serial number than the rest, and it also had a different Flash Type, so the older card had a serial starting with A1651 and had a GD25Q41B Flash Type, while the other 5 had a serial starting with A1701 and had a Flash Type M25P20/c.

Before flashing, all cards were running 24MH+ with stock settings.

After flashing, the A1651 card is running 27MH+, while the A1701 cards are running 29MH+

There was also a difference in the Device Manager recognition of the cards after flashing, where the A1651 card is detected as “AMD Radeon (TM) RX 480”, while the A1701 cards are detected as “Radeon (TM) RX 480 Graphics”.

All cards are running the driver 21.19.137.514 dated 21.11.2016

The power consumption from the wall went from 1270W before flash, to 980W after flash.

Everything is now running stable at 173MH+ with no additional OC manually.

I am trying to single out 1 specific card in a 6 card rig with different parameters than the rest, so let’s say:

-tt 70,75,75,75,75,75 -mclock 2100,2150,2150,2150,2150,2150

meaning in this case I want to set GPU 0 to 70C and 2100 memory clock, while the rest has 75C and 2150 clock.

The problem I am seeing is that GPU X is not necessarily the same card after every boot, so is there a way to force this, or figure it out during the start of Claymore, so you can adjust the same card every time?