[OS] rxOC easy-to-use Linux AMD Mining v_stopgap

[woodl1]

First all, Thank you very much for the awesome work you do!!

So,

Tried to put the DAG fix using the following:

#Install new ROCm Kernel:

wget -qO - http://repo.radeon.com/rocm/apt/debian/rocm.gpg.key | sudo apt-key add -
sudo sh -c 'echo deb [arch=amd64] http://repo.radeon.com/rocm/apt/debian/ xenial main > /etc/apt/sources.list.d/rocm.list'
sudo apt-get update
sudo apt-get install rocm

#edit grub
sudo nano  /etc/default/grub
#2MB fragments for Ellesmere are enabled with a grub option:
GRUB_CMDLINE_LINUX="amdgpu.vm_fragment_size=9"

sudo update-grub
sudo reboot


The problem is the first GPU will now do about half speed while the other GPU's are running properly faster. Tried multiple times, different gpu's, etc, same results.

I know it's being updated in V13, anybody have a clue how to fix this first GPU problem in the meantime? I've just reverted back to untouched V12.

If anyone cares, my 580's run at about 25(no oc), with dag fix 27, but the first gpu will run at 14, would be nice to get them all to 27.

What happens if you run headless?  Is the first GPU still nerfed?

I did the same to apply dagfix, with the same effect on the GPU0, tried to run it headless and yes, it is the same slow without monitor attached. By the way I had the same problem with nvOC headless P106 rigs, and it was easily solved by reconfiguring xorg with nvidia tool when no monitor attached. Maybe reconfiguration is the right way here too, but I don't know how to do it with RX cards.

[woodl1]

First all, Thank you very much for the awesome work you do!!

So,

Tried to put the DAG fix using the following:

#Install new ROCm Kernel:

wget -qO - http://repo.radeon.com/rocm/apt/debian/rocm.gpg.key | sudo apt-key add -
sudo sh -c 'echo deb [arch=amd64] http://repo.radeon.com/rocm/apt/debian/ xenial main > /etc/apt/sources.list.d/rocm.list'
sudo apt-get update
sudo apt-get install rocm

#edit grub
sudo nano  /etc/default/grub
#2MB fragments for Ellesmere are enabled with a grub option:
GRUB_CMDLINE_LINUX="amdgpu.vm_fragment_size=9"

sudo update-grub
sudo reboot


The problem is the first GPU will now do about half speed while the other GPU's are running properly faster. Tried multiple times, different gpu's, etc, same results.

I know it's being updated in V13, anybody have a clue how to fix this first GPU problem in the meantime? I've just reverted back to untouched V12.

If anyone cares, my 580's run at about 25(no oc), with dag fix 27, but the first gpu will run at 14, would be nice to get them all to 27.

What happens if you run headless?  Is the first GPU still nerfed?

I did the same to apply dagfix, with the same effect on the GPU0, tried to run it headless and yes, it is the same slow without monitor attached. By the way I had the same problem with nvOC headless P106 rigs, and it was easily solved by reconfiguring xorg with nvidia tool when no monitor attached. Maybe reconfiguration is the right way here too, but I don't know how to do it with RX cards.

Issue with slow GPU0 solved by setting it's PCIe slot to Gen2 in BIOS. I had Gen1 for all slots before.

[MentalNomad]

I've decided to stop waiting for v0013 and just run with v0012 for now because all my 1000-series rigs are now stable (thanks, @fullzero and all the contributors!) and my six Radeon RX470s were sitting idle (boo! BOO!)

But I'm running into a snag. On First Boot with the v0012 build after imaging (with the same solution I successfully use for imaging nvOC), the rig falls into BusyBox, as shown in the top screen, here:



Second boot goes into Grub, which allows me to start Ubuntu, but goes into the second image in the image above, reporting some sort of corruption and need for fsck.

I've scratched and re-imaged the USB key (same model used with nvOC), I've removed all video cards but the one in the primary X16 slot, I've pre-edited the oneBash file and dropped it into various folders for First Boot, but no joy.

It's on an Asus Z170-AR mobo - which worked fine in Win 10 for mining 6 cards after BIOS update. I have Above 4G enabled.

Any suggestions or direction for troubleshooting? I'm happy to provide whatever info anyone needs to assist. (Warning: I am still getting my bearings in Ubuntuland.)

I'm currently set up with a single video card in the primary X16 slot, connected via USB riser. On First Boot, I briefly get the purple Ubunu desktop background before the screen goes black and it drops into BusyBox.

If necessary, I'll pull the mobo out so that I can do a first boot with a card directly in the slot - that's something I've found necessary for Windows builds, but which I have not found to be necessary for nvOS builds.

Thanks, in advance!

[MentalNomad]

Update:

Figured out that I can run fsck manually with the following:

Code:

fsck /dev/sda2

Afterwards, CTRL-ALT-DEL gets Ubuntu to start, but no GUI... just a login.

Will continue trying to troubleshoot and look through the thread again to see how to start the GUI...

[MentalNomad]

OK, mined ETH all night!

Looking forwards to next version, but meanwhile, thanks for all you've done for us!

 
Next job: upgrade Claymore to 10.0 and tweak OC...

[TheNewEthlite]

First all, Thank you very much for the awesome work you do!!

So,

Tried to put the DAG fix using the following:

#Install new ROCm Kernel:

wget -qO - http://repo.radeon.com/rocm/apt/debian/rocm.gpg.key | sudo apt-key add -
sudo sh -c 'echo deb [arch=amd64] http://repo.radeon.com/rocm/apt/debian/ xenial main > /etc/apt/sources.list.d/rocm.list'
sudo apt-get update
sudo apt-get install rocm

#edit grub
sudo nano  /etc/default/grub
#2MB fragments for Ellesmere are enabled with a grub option:
GRUB_CMDLINE_LINUX="amdgpu.vm_fragment_size=9"

sudo update-grub
sudo reboot


The problem is the first GPU will now do about half speed while the other GPU's are running properly faster. Tried multiple times, different gpu's, etc, same results.

I know it's being updated in V13, anybody have a clue how to fix this first GPU problem in the meantime? I've just reverted back to untouched V12.

If anyone cares, my 580's run at about 25(no oc), with dag fix 27, but the first gpu will run at 14, would be nice to get them all to 27.

What happens if you run headless?  Is the first GPU still nerfed?

I did the same to apply dagfix, with the same effect on the GPU0, tried to run it headless and yes, it is the same slow without monitor attached. By the way I had the same problem with nvOC headless P106 rigs, and it was easily solved by reconfiguring xorg with nvidia tool when no monitor attached. Maybe reconfiguration is the right way here too, but I don't know how to do it with RX cards.

Issue with slow GPU0 solved by setting it's PCIe slot to Gen2 in BIOS. I had Gen1 for all slots before.

Thank you very much! This solved my issue :-)

[woodl1]

TheNewEthlite, thanks to alko67bi - he gave the idea

[burntone]

fullzero, when should we expect a v0013 on the rxOC?
I`m currently using nvOC and I`m very pleased with the new implementatinos of telegraph messages and the watchdog, great job!

[risinqsun]

thank you for v12.   v12 better than v11 seems and stabiliz.  I can download thx but google download link some slow you should upload mega or ftp .

[Zakyn]

Seems like there is new drivers for Nix.

Hoping for new version to come out so I can move on from Windows to RxOC. Those vegas are waiting :-P

https://support.amd.com/en-us/kb-articles/Pages/AMDGPU-Pro-Beta-Mining-Driver-for-Linux-Release-Notes.aspx

[shade123]

@fullzero
Hey man, I know this is kinda stupid and I'm sure I'm missing something, but I've searched and cannot find a solution, this occurs on both miners Genoil and Claymore, any thoughts?
Genoil :
Claymore : https://i.imgur.com/QIXKhxG.jpg
Genoil 1 : https://i.imgur.com/QmjFWjn.jpg
Genoil 2 : https://i.imgur.com/csX5vKq.jpg
Onebatch : https://pastebin.com/CqqYSfGd
Any help is appreciated! Thank you and have a nice day
Edit note : I tried a few different pools and they're all the same...

[MentalNomad]

@fullzero
Hey man, I know this is kinda stupid and I'm sure I'm missing something, but I've searched and cannot find a solution, this occurs on both miners Genoil and Claymore, any thoughts?
Genoil :
Claymore : https://i.imgur.com/QIXKhxG.jpg
Genoil 1 : https://i.imgur.com/QmjFWjn.jpg
Genoil 2 : https://i.imgur.com/csX5vKq.jpg
Onebatch : https://pastebin.com/CqqYSfGd
Any help is appreciated! Thank you and have a nice day
Edit note : I tried a few different pools and they're all the same...

Those are not miner errors and are not GPU errors.

Those are network errors.

Are you on WiFI? Try wired, instead.

From your mining rig, try pointing Firefox at a web site that tests the quality of your internet connection.

[shade123]

Those are not miner errors and are not GPU errors.

Those are network errors.

Are you on WiFI? Try wired, instead.

From your mining rig, try pointing Firefox at a web site that tests the quality of your internet connection.

@MentalNomad

Hey man thanks for replying, I'm using wired internet that is running at 40 Mbps/sec , I solved it by re-installing rxOC and trying a pool I haven't tried before, anyway, I have now another issue LOL, I just got a GTX 1070 and I wanna put it on my mining rig which is running rxOC and R9 380x but I only have a 32gb USB for my rig, I saw ocFACTORY which works for both but I only have 32 gb space

@FullZero

My motherboard is Gigabyte G1.Sniper B7 running rxOC with R9 380X at 17.8Mh/s non OC if you wanna add it to the list, I'm trying to get that GPU (R9 390X) and a GTX 1070 to work but I only have a 32GB USB and I cannot find anything that works except your rxOC... Do you have any idea? :/ Thank you!

[MentalNomad]

@shade123 - rxOC only has video drivers for AMD cards. @FullZero has a separate build/project called nvOC with NVIDIA drivers.

nvOC is actually much better developed:

https://bitcointalk.org/index.php?topic=1854250.0

[shade123]

@shade123 - rxOC only has video drivers for AMD cards. @FullZero has a separate build/project called nvOC with NVIDIA drivers.

nvOC is actually much better developed:

https://bitcointalk.org/index.php?topic=1854250.0

I read about it but I only have one rig and I have an AMD R9 380X and GTX 1070...
I wanted to download factoryOC which is another OS/build that works for both AMD and Nvidia but I only have 32 GB of space and the main problem is his the factoryOC download links are not working idk why...

https://bitcointalk.org/index.php?topic=1996684

[MentalNomad]

@shade123 - rxOC only has video drivers for AMD cards. @FullZero has a separate build/project called nvOC with NVIDIA drivers.

nvOC is actually much better developed:

https://bitcointalk.org/index.php?topic=1854250.0

I read about it but I only have one rig and I have an AMD R9 380X and GTX 1070...
I wanted to download factoryOC which is another OS/build that works for both AMD and Nvidia but I only have 32 GB of space and the main problem is his the factoryOC download links are not working idk why...

https://bitcointalk.org/index.php?topic=1996684

You can try running nvOC and then adding AMD driver for your 380x... but the automated setting of power and OC is effective on NVidia, but not on AMD right now.

(Under Linux. Under Windows, AMD tweaking works better. It's a function of the drivers available, so no good solution for Linux, right now. I think that's part of why rxOC gets less attention - until drivers make tweaking more possible, there's not as much demand for rxOC.)

[fullzero]

First all, Thank you very much for the awesome work you do!!

So,

Tried to put the DAG fix using the following:

#Install new ROCm Kernel:

wget -qO - http://repo.radeon.com/rocm/apt/debian/rocm.gpg.key | sudo apt-key add -
sudo sh -c 'echo deb [arch=amd64] http://repo.radeon.com/rocm/apt/debian/ xenial main > /etc/apt/sources.list.d/rocm.list'
sudo apt-get update
sudo apt-get install rocm

#edit grub
sudo nano  /etc/default/grub
#2MB fragments for Ellesmere are enabled with a grub option:
GRUB_CMDLINE_LINUX="amdgpu.vm_fragment_size=9"

sudo update-grub
sudo reboot


The problem is the first GPU will now do about half speed while the other GPU's are running properly faster. Tried multiple times, different gpu's, etc, same results.

I know it's being updated in V13, anybody have a clue how to fix this first GPU problem in the meantime? I've just reverted back to untouched V12.

If anyone cares, my 580's run at about 25(no oc), with dag fix 27, but the first gpu will run at 14, would be nice to get them all to 27.

What happens if you run headless?  Is the first GPU still nerfed?

I did the same to apply dagfix, with the same effect on the GPU0, tried to run it headless and yes, it is the same slow without monitor attached. By the way I had the same problem with nvOC headless P106 rigs, and it was easily solved by reconfiguring xorg with nvidia tool when no monitor attached. Maybe reconfiguration is the right way here too, but I don't know how to do it with RX cards.

Issue with slow GPU0 solved by setting it's PCIe slot to Gen2 in BIOS. I had Gen1 for all slots before.

Good to know.  I have a way of jamming the system to patch this: but if setting gen2 works it is the much better solution. 

[fullzero]

I've decided to stop waiting for v0013 and just run with v0012 for now because all my 1000-series rigs are now stable (thanks, @fullzero and all the contributors!) and my six Radeon RX470s were sitting idle (boo! BOO!)

But I'm running into a snag. On First Boot with the v0012 build after imaging (with the same solution I successfully use for imaging nvOC), the rig falls into BusyBox, as shown in the top screen, here:



Second boot goes into Grub, which allows me to start Ubuntu, but goes into the second image in the image above, reporting some sort of corruption and need for fsck.

I've scratched and re-imaged the USB key (same model used with nvOC), I've removed all video cards but the one in the primary X16 slot, I've pre-edited the oneBash file and dropped it into various folders for First Boot, but no joy.

It's on an Asus Z170-AR mobo - which worked fine in Win 10 for mining 6 cards after BIOS update. I have Above 4G enabled.

Any suggestions or direction for troubleshooting? I'm happy to provide whatever info anyone needs to assist. (Warning: I am still getting my bearings in Ubuntuland.)

I'm currently set up with a single video card in the primary X16 slot, connected via USB riser. On First Boot, I briefly get the purple Ubunu desktop background before the screen goes black and it drops into BusyBox.

If necessary, I'll pull the mobo out so that I can do a first boot with a card directly in the slot - that's something I've found necessary for Windows builds, but which I have not found to be necessary for nvOS builds.

Thanks, in advance!

Update:

Figured out that I can run fsck manually with the following:

Code:

fsck /dev/sda2

Afterwards, CTRL-ALT-DEL gets Ubuntu to start, but no GUI... just a login.

Will continue trying to troubleshoot and look through the thread again to see how to start the GUI...

OK, mined ETH all night!

Looking forwards to next version, but meanwhile, thanks for all you've done for us!

 
Next job: upgrade Claymore to 10.0 and tweak OC...

I have spent a few hours messing around with the new driver and have a version I am using as a sort of stopgap to test the new driver.  With most GPUs it does have better Ethash peformance, but Equihash seems the same in my tests so far. 

[fullzero]

fullzero, when should we expect a v0013 on the rxOC?
I`m currently using nvOC and I`m very pleased with the new implementatinos of telegraph messages and the watchdog, great job!

I am going to push a -1.4 update to nvOC then continue porting the features over to the amd api.

This would be a lot faster, but members keep adding and modifying the nvOC features which I want to stabilize / standardize before I port them.  more of a short term loss long term gain.

[fullzero]

thank you for v12.   v12 better than v11 seems and stabiliz.  I can download thx but google download link some slow you should upload mega or ftp .

I will add a mega link for the next version.