Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux)

[malandante]

UPDATE: Fixed bug with Linux build when miner is closed after some time, re-download v4.1 for Linux.

Testing again

I have been testing overnight and there is another problem. The miners don't stop mining, shares are still coming on both eth and dcr but this issue appears:

read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
...

Very fast scrolling, and cpu usage skyrockets.

Looks like another bug with porting sockets to Linux. Today I will check/fix it.

Great! Good job, Claymore!

[1714108071]

1714108071

[1714108071]

1714108071

[1714108071]

1714108071

[Claymore]

- UPDATE: Fixed bug with Linux build when miner shows message "read(): Resource temporarily unavailable", re-download v4.1 for Linux.

[wariner]

Hi Claymore and thank you for the miner,

on my pc with ubuntu 14.04 the current consumption increases by approximately about 20% (and temperature) respect ethminer, if I use "mode 1" (only eth) and "-ethi 6" the consumtpion it's similar.

This is it normal? or it's a linux problem?

[ilgio]

Hi

True wariner
It's increase the consumation and temperature to me

[Claymore]

Hi Claymore and thank you for the miner,

on my pc with ubuntu 14.04 the current consumption increases by approximately about 20% (and temperature) respect ethminer, if I use "mode 1" (only eth) and "-ethi 6" the consumtpion it's similar.

This is it normal? or it's a linux problem?

If you mine Ethereum+Decred the power consumption will be higher than if you mine Ethereum only.

[Gaglam]

Testet it over the last night.
Made no difference for me at dwarfpool. Switched back to ethminer.
I'm on Ubuntu with 4x R9 280x.

Any suggestions how to improve?
Also tried -ethi -16384 and it just got slower.

[Claymore]

Testet it over the last night.
Made no difference for me at dwarfpool. Switched back to ethminer.
I'm on Ubuntu with 4x R9 280x.

Any suggestions how to improve?
Also tried -ethi -16384 and it just got slower.

try set "-etha" to "0" and then to "1". From readme:
.....
-etha   Ethereum algorithm mode. 0 - optimized for fast cards, 1 - optimized for slow cards. -1 - autodetect (default).
   You can also set this option for every card individually, for example "-etha 0,1,0".
.....

If it does not help - make sure you use Catalyst 15.12. If still see higher hashrate on pool with ethminer - use it.

PS. For Dual mode: I get the best hashrate on Ethereum if I change Decred intensity manually by "+" and "-" keys at some value (press "s" to see current stats).
After that you can set best found value with "-dcri" option.

[Mrga666]

Claymore help, I am getting way too much of fluctuation in my effective hash rate in 4.1 miner for windows. Miner shows 62.4MH/s but effective is 55 MH/s, and it goes down to 40 and up to 65 etc. With 3.2 miner I got way better effective hash rate. Help?

I got -ethi 16 and ethermine pool.

[citronick]

Claymore - do we need to put "--no-precompute" in batch file to ensure DAG files are current? Just curious.
I hardly see any rejects mining at ethpool, so surely DAGs are ok.
I have read somewhere that a guy got no rewards because he was mining with a wrong DAG. Thanks

[Claymore]

Claymore help, I am getting way too much of fluctuation in my effective hash rate in 4.1 miner for windows. Miner shows 62.4MH/s but effective is 55 MH/s, and it goes down to 40 and up to 65 etc. With 3.2 miner I got way better effective hash rate. Help?

I got -ethi 16 and ethermine pool.

With your hashrate (<100MH) you should run miner for at least 24 hours to see average hashrate. Run 3.2 and 4.1 for 24 hours and compare average hashrates, they must be the same. Checking miner for a few hours with not very high hashrate is not a good idea. If you still see that 3.2 works much faster: I did not change OpenCL code between these versions, so it must be some kind of magic I think.

[Claymore]

Claymore - do we need to put "--no-precompute" in batch file to ensure DAG files are current? Just curious.
I hardly see any rejects mining at ethpool, so surely DAGs are ok.
I have read somewhere that a guy got no rewards because he was mining with a wrong DAG. Thanks

Do you see "--no-precompute" option in readme file? I don't see So it is not supported.
DAGs are often corrupted when they are stored as files. But now miner generates DAGs on GPU so it can be corrupted only if your GPU does not work properly. I did not see such behaviour yet. If I get some feedbacks about it, I can add CRC check for DAGs (so GPU will check CRC after DAG generation).

[citronick]

Claymore - do we need to put "--no-precompute" in batch file to ensure DAG files are current? Just curious.
I hardly see any rejects mining at ethpool, so surely DAGs are ok.
I have read somewhere that a guy got no rewards because he was mining with a wrong DAG. Thanks

Do you see "--no-precompute" option in readme file? I don't see So it is not supported.
DAGs are often corrupted when they are stored as files. But now miner generates DAGs on GPU so it can be corrupted only if your GPU does not work properly. I did not see such behaviour yet. If I get some feedbacks about it, I can add CRC check for DAGs (so GPU will check CRC after DAG generation).

thanks - that makes sense - my shares are accepted since day 1 using the dual miner, so I guess my cards are functioning well.

however, this case about someone mining on a wrong DAG (not corrupted) but maybe a past version -- I wonder how does a case like that happen?

[osnwt]

It's also a good idea to read RTFM for pool. I will update sample for dwarfpool, you are right, current sample is not good if you want to see hashrate.
I had to add "-eworker" because there are 3 stratum versions and many pools that also add some variants on these versions for logon process. Some pools ignore "-worker" but search worker name in wallet string, some require only wallet as username but require worker name when worker sends a share. Some work with both approaches but don't show hashrate as expected. Such problems are not related to the miner, they are because of some chaos in Ethereum stratum implementations on pools.

Actually, I did RTFM for pools I used. But it was never mentioned in the README for the miner that wallet option can be somehow combined with something else to build up the final URL. Being it called 'url' option - no questions. But I expected that if you mention ethermine and dwarfpool in your examples and provide 3 different stratum modes for them, then their specifics are already implemented by -esm option.

Anyway, I agreed that it is not the issue with the miner, but with the documentation. Thank you for great product, I like it and use it now.

[Claymore]

Claymore - do we need to put "--no-precompute" in batch file to ensure DAG files are current? Just curious.
I hardly see any rejects mining at ethpool, so surely DAGs are ok.
I have read somewhere that a guy got no rewards because he was mining with a wrong DAG. Thanks

Do you see "--no-precompute" option in readme file? I don't see So it is not supported.
DAGs are often corrupted when they are stored as files. But now miner generates DAGs on GPU so it can be corrupted only if your GPU does not work properly. I did not see such behaviour yet. If I get some feedbacks about it, I can add CRC check for DAGs (so GPU will check CRC after DAG generation).

thanks - that makes sense - my shares are accepted since day 1 using the dual miner, so I guess my cards are functioning well.

however, this case about someone mining on a wrong DAG (not corrupted) but maybe a past version -- I wonder how does a case like that happen?

Miner will show you warnings that GPU finds incorrect shares again and again. Such incorrect shares won't be sent to pool.

[Gaglam]

Testet it over the last night.
Made no difference for me at dwarfpool. Switched back to ethminer.
I'm on Ubuntu with 4x R9 280x.

Any suggestions how to improve?
Also tried -ethi -16384 and it just got slower.

try set "-etha" to "0" and then to "1". From readme:
.....
-etha   Ethereum algorithm mode. 0 - optimized for fast cards, 1 - optimized for slow cards. -1 - autodetect (default).
   You can also set this option for every card individually, for example "-etha 0,1,0".
.....

If it does not help - make sure you use Catalyst 15.12. If still see higher hashrate on pool with ethminer - use it.

PS. For Dual mode: I get the best hashrate on Ethereum if I change Decred intensity manually by "+" and "-" keys at some value (press "s" to see current stats).
After that you can set best found value with "-dcri" option.

I think i'm still on catalyst 15.9?

display: :0  screen: 0
OpenGL vendor string: Advanced Micro Devices, Inc.
OpenGL renderer string: AMD Radeon R9 200 Series
OpenGL version string: 4.5.13399 Compatibility Profile Context 15.201.1151

[Pustul]

Hi Claymore,

Sorry if I missed it, but do you plan on integrating some sort of API that will send stats to an IP address so that I can build a little webpage to check on my miners?

Thank you

[malandante]

Claymore, In certain cases after automatic respawn of the miner, the DAG can't be loaded error appears (I don't remember the exact words). I'm talking about ubuntu 15.04 with catalyst 15.12. It is an error that used to happen with the stock ethminer and the only solution was to reboot the computer. Is it possible to make the watchdog flush the GPU ram before respawning the miner?

[Claymore]

Claymore, In certain cases after automatic respawn of the miner, the DAG can't be loaded error appears (I don't remember the exact words). I'm talking about ubuntu 15.04 with catalyst 15.12. It is an error that used to happen with the stock ethminer and the only solution was to reboot the computer. Is it possible to make the watchdog flush the GPU ram before respawning the miner?

The only option is to reboot computer. Miner must execute "reboot.bash" if you use "-r 1", though I did not test it in Linux. If it does not work, send me sample "reboot.bash" that must reboot computer and I will check it.

[Claymore]

Hi Claymore,

Sorry if I missed it, but do you plan on integrating some sort of API that will send stats to an IP address so that I can build a little webpage to check on my miners?

Thank you

It will work in a different way: in a few days I will release utility-proxy that will monitor rigs and will have web server that displays status.

[alexcrys]

Hi Claymore,

It is possible to work on miningrigrentals ?... not working thx