Download AMDGPU-Pro Driver Version 16.30 released by AMD.
The basics of the install:
1. download amdgpu-pro_16.30.3-306809.tar.xz
2. extract the compressed file to your home folder or somewhere.
3. goto that folder and now run the installation script "sudo ./amdgpu-pro-install"
AMD's page
support.amd.com/en-us/kb-articles/Pages/AMD-Radeon-GPU-PRO-Linux-Beta-Driver%E2%80%93Release-Notes.aspxAnd here is the direct link to the compressed file:
https://www2.ati.com/drivers/linux/amdgpu-pro_16.30.3-306809.tar.xzNOTE: You have to extract the compressed file and then run the script to install the driver to make opencl devices detectable (for ethminer, oclhashcat, etc.)
I am using 6 - RX 480 (4GB model ) in Ubuntu 16.04 LTS
Stock settings. I believe the ram is at 7000mhz vs 8000 mhz in 8GB version.
Preliminary benchmarks Ubuntu 16.04 LTS
Driver is 16.30 for Linux
Genoil 1.1.7
2 - XFX 4gb rx 480
4 - Sapphire 4gb rx 480
Doing about 134-139 mh/s. I'd say its doing about 22.5 or 23mh/s in Ubuntu 16.04 in stock mode.
Next up is finding out how to overclock/undervolt GPUs in Ubuntu 16.04 with atitweak or something else (amdconfig/aticonfig do not work anymore unless I need to install the SDK.. we shall see.)
Comments
I use one SAPPHIRE 8G RX480
dmesg show :
amdgpu 0000:01:00.0: GPU fault detected: 147 0x0bd00401
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x0645DF74
amdgpu 0000:01:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0E0C4001
edit all went ok then when i try mining it freezes the computer gets as far as generating one big chunck or it will start mining with large fluctuation in Mh/s then soon freeze.
As usual, adjusting voltages in Linux isn't possible - you have to flash the vbios. Beyond that, it seems like there is currently a bug affecting setting at least the memory clock (pp_mclk_od): "VDDCI is larger than max VDDCI in VDDCI Voltage Table!" error in syslog regardless of the value set (0-20), and the memory drops to the low-state clock speed.
Phoronix seems like the best source for the latest info about AMD GPUs on Linux, and they reported similar findings with the pp_mclk_od bug @ http://www.phoronix.com/scan.php?page=article&item=amd-rx480-od&num=1
all went ok then when i try mining it freezes the computer gets as far as generating one big chunck or it will start mining with large fluctuation in Mh/s then soon freeze.
Tried a few things in ubuntu and various switches but no joy, thought it might be the card. Installed win10 and it's mining away at 20-21Mh/s on stock setting 82°C. Have some Sapphire 480's on order will try those on ubuntu.
Edit:
I've heard a few ppl say these run hot so combined with OS limitation, I might rethink about using these for mining and stick with the power hungry 390's
I get the exact same error in the logs (dmesg) running ethminer. I haven't found a solution. In addition to that it uses up about 1+GB of additional disk space about every 5 minutes or so. And that is with ethminer logging level set to just -v 3 (default level is higher at 8).
It still hashes, just not very good with an average of 16Mh/s.
ASUS 8GB RX480
Ubuntu 16.04 LTS
AMDGPU-PRO v16.3
parity v1.2.2-beta
ethminer v1.2.9
[ 1084.587016] VM fault (0x01, vmid 3) at page 47326788, read from 'TC0' (0x54433000) (8) [ 1084.587016] amdgpu 0000:07:00.0: GPU fault detected: 147 0x05708801 [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x00E0983C [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x06088002 [ 1084.587016] VM fault (0x02, vmid 3) at page 14719036, read from 'TC6' (0x54433600) (136) [ 1084.587016] amdgpu 0000:07:00.0: GPU fault detected: 147 0x0d78c401 [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x03B9780A [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x060C4001 [ 1084.587016] VM fault (0x01, vmid 3) at page 62486538, read from 'TC3' (0x54433300) (196) [ 1084.587016] amdgpu 0000:07:00.0: GPU fault detected: 147 0x0c88c801 [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x016D8DF6 [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x06044001 [ 1084.587016] VM fault (0x01, vmid 3) at page 23956982, read from 'TC5' (0x54433500) (68) [ 1084.587016] amdgpu 0000:07:00.0: GPU fault detected: 147 0x09e08801 [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x006F7D1B [ 1084.587016] amdgpu 0000:07:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x060C4002 etc. etc...
According to Phoronix's tests (https://phoronix.com/scan.php?page=news_item&px=AMDGPU-Mesa-RX480-July-16) it is giving better performance on some games compared to AMDGPU-PRO.
I've successfully reduced power draw by 20-25 watts at the wall outlet.
Stock Total System power draw was 205 watts/hr @ the wall with 1 - RX 480 card in.
Total System power draw is now 180-185 watt/hr @ the wall
Mining 22.5 to 23 mh/s on a 4gb Sapphire rx 480 in Ubuntu version 16.04 LTS.
In Windows the card would probably run 24 to 25mh/s.
I believe there is a bit more room to go. Gonna test a bit more then probably take a break as its nice outside
I'd love to run at least 4.7rc3 because that's where the patches to properly set memory clocks came into the kernel tree. Alternatively, I was considering compiling a kernel with that specific patch backported into kernel 4.6.4.
That is with the "110" branch that compiled itself to a version 1.1.7.
Now to figure out/learn how to adjust settings to increase the hash/power ratio.
Thanks again!
On another note I've seen some guy in this forum that sells replacment mod BIOS'es (for R9 nano). What is that all about? Is he creating his own chips that somehow make mining better?
However now that you say this, is this what @GreaterNinja is probably doing with his RX 480 and then moving his card back to the Ubuntu OS?
And yes, that's likely what 'Ninja is doing. That's what I do currently with my Nanos.
Hopefully AMD removes the signature checks later like with hawaii.
Both @dlehenky and @work said it all
Thx guys
(will be more active again in a week or so.. very very busy lately..)